ATTACHMENT RND CHARGE I/i PROCESS DEPT OF ELECTRICAL …

tAD-A162 747 ELECTRON PRODUCTION ELECTRON ATTACHMENT RND CHARGE I/iRECOMBINATION PROCESS I (U) SAN DIEGO STATE UNIV CADEPT OF ELECTRICAL AND COMPUTER ENGIN L C LEE

UNCLASSIFIED 10 SEP 35 AFOSR-TR--85-li32 AFOSR-82-0314 F/G 7/4 N

EEEEEEEEEoi

I lffll...f~f.

JjW

11i11 396 2 132 .

1.5 LA .

MICROCOPY RESOLUTION TEST CHART

NATIONAL BUREAU OF STANDARDS-1963-A

414

-. ~ ~~ ~ 5~ 1-~~I~~,I

September 10, 1985

0 Annual Scientific Report for Grant No. AFOSR-82-0314Covering the Period from 1 August 1984 to 31 July 1985

< ELECTRON PRODUCTION, ELECTRON ATTACHMENT, AND CHARGE4 RECOMBINATION PROCESS IN HIGH PRESSURE GAS DISCHARGES

By: Long C. LeeDepartment of Electrical and Computer EngineeringSan Diego State University DTICSan Diego, California 92182-0190Telephone: (619) 265-3701 S D-EC 30 l 0

P ~Prepared for: lp

U. S. Air Force Office of Scientific ResearchBolling Air Force BaseWashington, D. C. 20332-6448Attention: Major Bruce L. Smith

Directorate of Physical and GeophysicalSciences -

Lai N ATE1

85 12 30 039

SECURITY LASIFICA T10IO, N o ("2e brnEnrered).i~ ~ ~~ G i I 'L 'I 7t "

READ INSTRUYCTIONSREPORT DOCuMENTA.TIION PAGE RA TUTOBEFORE COMPLETING FORM

I. REPORT NUMBER 2" GOVT ACCES P7I S CATALOG NUMBER

4. TITLE (and Subtitle) S. TYPE OF REPORT & PERIOD COVERED

Electron Production, Electron Attachment, and Annual Report

Charge Recombination Process in High Pressure 1 August 1984 - 31 July 1985

Gas Discharges 6. PERFORMING OG. REPORT NUMBER

7. AUTHOR(&) S. CONTRACT OR GRANT NUMBER,.:

Long C. Lee AFOSR-82-0314

9. PERFORMING ORGANIZATION NAME AND ADDRESS 10. PROGRAM ELEMENT. PROJECT. TASK

AREA &1 WO RK NUMBERS

San Diego State University (

San Diego, California 92182

11. CONTROLLING OFFICE NAME AND ADDRESS 1'. REPOTDATE

U. S. Air Force of Scientific Research/ 10 September 1985

Bolling Air Force Base 13. NUMBER OF PAGES

14. MONI, TOR G AdENCY NAME -FADDRESS(f different from Controlling Office) IS. SECURITY CLASS. fo! !his report,

Unclassified

ISa. DECLASSIFICATION DOWNGRADINGSCHEDULE

16. DISTRIBUTION STATEMENT (of this Report)

Approved for public release; distribution unlimited

*7. DISTRIBUTION STATEMENT (of the abstract entered in Block 20, if different from Report)

IS. SUPPLEMENTARY NOTES

The findings in this report are not to be constructed as an 4ficial position

of the Department of the Air Force, unless so designated by other authorized

documents.

19. KEY WORDS (Continue on reverse side if necessary and identify by block number)

-Electron production; electron attachment' electron diffusion; charge recombina-

tion; electron conduction current; temporary negative ion; space charge effect;

plasma decay; electron leakage from plasma; electron attaching gas; excimer

laser; parallel-plate drif -tupe apparatus; computer modeling .

2 ABSTRACT (Continue on reverse side If neeesaary and identify by block number)

)The electron productions from two-photon-ionization of CS2 , SO2 and (CH )3N

at 193 nm were investigated and their coefficients were measured. The effect

of space charge on the electron conduction pulse was observed as a function

of charge density and the applied electric field. The electron attachment rate

constants of H20 and C F8 in buffer gases of Ar, N2 , and CH4 were measured. The

electron attachment rate constants for the gas mixtures of H2 0-Ar, C3F8 -N2 and

C3F8-CH4 increase with increasing E/N. These characteristics are useful for the

application of opening switchef. The electron drift velocitJe5 . r±- gas

I NOV. 65 S BSLET IpaY

DD JAN 73 1473 EDITION OF I NOV 65 IS OBSOLETE

SECURIT;tW 1*b,?TWSh PKGC_(Whn Dat& Entoead,

ere,-

AIR "RC3 " ( .....

I. INTRODUCTION

This annual report covers the period from August 1, 1984 to

July 31, 1985 for the research project under Grant No. AFOSR-82-

0314. In this research program, the processes for electron

production, electron attachment and charge recombination in high

pressure gaseous mixtures were investigated. The information

obtained from this research is currently needed for developing

various electrical switching devices as well as for understanding

the basic phenomena in plasma physics.

Various electrical switching devices, such as high energy

and high repetition-rate discharge switches, opening switches,

radiation or e-beam controlled switches, are needed for the

development of high power lasers, fusion experiment, magnetic

energy storage system, as well as particle beam experiment. High

pressure gaseous discharges are frequently used in these

switching devices. For a specific discharge switch, it may

require some special characterisitcs pertinent to the pulse rise

and decay times, discharge stability, dishcarge uniformity, and

current density. These characterisitcs depend strongly on

electron transport parameters, such as electron drift velocity,

electron attachment, detachment and ionization coefficients as

well as charge recombination rates. The electron transport

parameters of various gases are thus needed for designing various

discharge switches, and they are measured in this research

program.

In addition to pracitcal applications, this research program

also developes new methods for the study of basic phenomena in

2

plasma physics. A high-charge-density plasma can be produced by

an intense excimer laser pulse by the multiphoton-ionization

process, and it can be used for examining the probability of

electron leakage from plasma and the decay time of plasma in high

gas pressures.

II. REASEARCH ACCOMPLISHMENTS

A parallel-plate drift-tube apparatus was used to

investigate electron transport parameters in high pressure

buffer gases. Electrons were initially produced either by

irradiation of the cathode or by two-photon-ionization of a trace

of trimethylamine in a high-pressure buffer gas with an intense

excimer laser pulse. An electric field produced by applying a

negative high voltage on the cathode was used to drift the

electrons. The transient voltage pulses induced by electron

motion between the electrodes were observed. The electron

attachment rates were obtained from the ratio of the amplitudes

of transient pulses with and without a gas attacher in buffer

gas. The electron drift velocity and the electron diffusion

coefficient were obtained by analyzing the time profile of

transient pulses.

In this period, the two-photon-ionization coefficients of

CS2 and S02 at 193 nm were also measured. These coefficients are

needed for determining the initial electron densities produced by

laser-ionization of trace amounts of CS2 and S02 in buffer gases.

Results accomplished in this period are described more

specifically below.

3

4.1.

1. Electron Production from Photoionization of CS9. S09 and

(CHr) gN

The electron productions from photoionization of CS 2 , S02,

and (CH3)3N in buffer gases of N2 and CH4 with an ArF excimer

laser (193 nm) were investigated. At low laser power and low gas

pressure, the number of electrons produced are proportional to

the square of laser power and the density of CS2, SO2 or (CH3 )3N,

indicating that electrons are produced by two-photon-ionization

(TPI) process. The TPI coefficients for CS 2 , S02 and (CH3)3N

were measured to be 3.7xl0 - 2 7 , 8.3xi0 -3 0 , and 1.7x10 - 2 7 cm 4 /W,

respectively. The results are described in more detail in a

paper attached in this report as Appendix A. This paper has been

accepted by the Journal of Applied Physics for publication.

2. Space Charae Effect on Electron Transport

The probability for electron leakage from a plasma depends

on the charge density and the applied field. The electron

leakage probability decreases with charge density and increases

with applied field. In an earlier observation of (CH3 )3N, we

find that the probability is a function of the ratio of charge

density to the applied field. This phenomenon is repeatly

observed in the case of CS2 which is briefly discussed in the

paper attached as Appendix A.

The electron conduction pulse is greatly affected by the

space charge effect. At high charge density, the electric field

induced by the space charge can over-suppress the applied

external field such that the conduction electrons move toward the

cathode, and the electron conduction current reverses its

4

direction as shown in Fig. 9 in Appendix A. This new phenomenon

is not expected from common belief that electrons should not move

toward the cathode. However, this phenomenon has been observed

in both cases of CS2 and (CH3)3N whenever the electron density is

high enough. We thus believe that this is a genuine physical

phenomenon, although the theoretical explanation for this

phenomenon is still not established. To understand this

phenomena, theoretical modeling of the charge density and the

time-dependent electric field is needed. We will continue to

pursue this problem in the next funding period.

3. Electron Attachment Coefficients of H20

The electron attachment rate constants of H20 in buffer

gases of Ar, N2 and CH4 were measured as a function of E/N. The

electron attachment rate constant of H20 in Ar increases with E/N

from 2 to 15 Td. This characteristic is desirable for the design

of opening swithces. For the H2 0-N2 and H20-CH4 gas mixtures,

the electron attachment to H20 is due to the formation of

"temporary" negative ions, whose lifetime is about 200 ns. The

foramtion of "temporary" negative ion has a certain effect on the

electron drift velocity. The electron drift velocities of H20 in

various buffer gases were measured in this experiment.

The results for the electron attachment of H20 in buffer

gases are reported in more detail in a paper attached as Appendix

B, which has been published in the Journal of Applied Physics.

4. Electron Attachment Coefficients of CqFA

The electron attachment coefficients of C3F8 in buffer gases

of Ar, N2, and CH 4 were measured as a function of E/N. The

5

electron attachment rate constants of C3F8 in N2 and CR 4 increase

with E/N, which are desirable for the design of opening switches.

The electron drift velocity in the C3FS-CH4 mixtures increases

with increasing E/N, and it reaches a peak and then decreases at

high E/N. The peak drift velocity shifts to high E/N as the C3F8

concentration increased. This characteristic satisfies the need

for the design of opening swtiches.

The results for the shortening of electron conduction pulses

by C3 FS are described in more detail in a paper attached as

Appendix C, which has been published in the Journal of Applied

Physics.

4 ~Ccesion For V41i NTIS CRA&IDTIC TABUnannounced

Jutfitlcon

ByDist. IUtaln I ale

Availability Codes *NSPtCEO

DISt AXvil "anefjor4peCial

6

. . . .. . . . . . . . . - . ,. -.. . . . + . . . , % , . , . 1.' , , -. , u"1r' ', . '- I ... l ,i+. , 4, j ..+ +- + t ++

III. CUMULATIVE PUBLICATIONS AND PRESENTATIONS

1. "Two-Photon-Ionization Coefficients of Propane, 1-Butene,Methylamines", L. C. Lee and W. K. Bischel, presented at the12th International Conference on the Physics of Electronicand Atomic Collisions, Gatlinburg, TN, July 15-21, 1981.

2. "Electron Attachment and Charge Recombination Following Two-Photon-Ionization of Methylamines", L. C. Lee and W. K.Bischel, presented at the 34th Gaseous ElectronicsConference, Boston, MA, October 20-23, 1981.

3. "Two-Photon-Ionization Coefficients of Propane, 1-Butene,and Methylamines", L. C. Lee and W. K. Bischel, J. Appl.Phys. 53Q. 203 (1982).

4. "Electron Ionization and Attachment Processes in DiffuseDischarges", L. C. Lee, presented to the Workshop on OpticalControl of Diffuse Discharges, Eugene, OR, December 2-3,1982.

5. "Diffusion of Electrons in Gases Under Electric Field", F.Li and L. C. Lee, presented at the 1983 Annual Meeting, APSDivision of Electronic and Atomic Physics, Boulder, CO, May23-25, 1983.

6. "Electron Longitudinal Diffusion Coefficients in Ar", F. Liand L. C. Lee, presented at the 36th Gaseous ElectronicsConference, Albany, NY, October 10-14, 1983.

a

7. "Space Charge Effect on the Electron Kinetics Occurring inAtmospheric Gas Pressure", L. C. Lee and F. Li, presented atthe 1984 IEEE International Conference on Plasma Science,St. Louis, Missouri, May 14-16, 1984.

8. "Shortening of Electron Conduction Pulses by ElectronAttachers 02, N20 and CF 4 ", L. C. Lee and F. Li, J. Appl.Phys., 56 3169 (1984).

9. "Space Charge Effect on Electron Conduction Current", L. C.Lee and F. Li, paper submitted to IEEE Transactions onPlamsa Science, 1984.

10. "Electron Attachment to C3F 8 in High Pressure Buffer Gases"W. C. Wang and L. C. Lee, presented at the DEAP Meeting,American Physical Society, Norman, Oklahoma, May 29-31,1985.

11. "Electron Attachment to H20 in Ar, N2, and CH 4 in ElectricField", W. C. Wang and L. C. Lee, J. Appl. Phys. 57 4360(1985).

7

12. "Shortening of Electron Conduction Pulses by ElectronAttacher C3F8 in Ar, N2 , and CH 4 ", W. C. Wang and L. C. Lee,J. Appl. Phys. 5, 184 (1985).

13. "Shortening of Electron Conduction Pulses by ElectronAttachers", L. C. Lee and W. C. Wang, presented at the IEEEPulsed Power Conference, Arlington, VA, June 10-12, 1985.

* 14. "Electron Attachment to H20, S02, and C3F8 in Ar, N2, andCH4" W. C. Wang, M. A. Fineman and L. C. Lee at XIVInternational Conference on the Physics of Electronic andAtomic Collisions, Palo Alto, CA, July 24-30, 1985.

15. "Electron Production by Photoionization of CS2 and S0 2 at193 nm" L. C. Lee and W. C. Wang, Accepted for presentationat the Thirty-Eighth Annual Gaseous Electronics Conference,Monterey, CA, Oct. 15-18, 1985.

16. "Electron Attachment to S02 and CS2 in Ar, N2, and CH4", W.C. Wang and L. C. Lee, Accepted for presentation at theThirty-Eighth Annual Gaseous Electronics Conference,Monterey, CA, Oct. 15-18, 1985.

17. "Two-Photon-Ionization Coefficients of CS2, S02 and(CH3)3N", W. C. Wang and L. C. Lee, paper accepted forpublication in J. Appl. Phys. (1985).

i8

IV. PERSONNEL INVOLVgD IN THIS RESEARCH

1. Principal Investigator:

Dr. Long C. Lee, Professor of Electrical & ComputerEnigneering

* 2. Dr. Morton A. Fineman, visiting Professor of Physics

3. Research Associates:

Dr. J. B. NeeDr. W. C. Wang

4. Students:

Mr. Chris FaganMr. Mike HomMr. Robert TahimicMr. F. MehranMs. M. Waxman

V. INTERACTIONS

1. We have constantly sent our papers to Dr. A. H. Guenther atthe Air Force Weapons Laboratory, Dr. Alan Garscadden at theAir Force Wright Aeronautical Laboratory, Dr. M. Gundersen atthe University of Southern California, Dr. J. T. Moseley atthe University of Oregon, Dr. Bob Reinovsky at the Air ForceWeapons Laboratory, Dr. S. K. Srivastava at the JetPropulsion Laboratory, and Dr. K. Schoenbach at the TexasTech University. We appreciate the useful comments and

Ssuggestions received from them.

presented at (i) the DEAP Meeting, Norman, Oklahoma May 29-31, 1985; (ii) the IEEE Pulsed Power Conference, Arlington,VA, June 10-12, 1985; (iii) the XIV International Conferenceon the Physics of Electronic and Atom Collisions, Palo Alto,CA, July 24-30, 1985.

3. We had visited the Jet Propulsion Laboratory on Nov. 30, 1984to discuss our results for the electron attachments to H20and C3 F8 with Dr. S. K. Srivastava and Dr. A. Chutjian.

-I .. . ... 1

Appendix A

"Two-Photon-Ionization Coefficients ofCS2, S02, and (CH3)3 N"

rF

---- - *

- Two-photon-ionization coefficients of CS2 , SO2 , and (CH 3)3NW.C. Wang and L. C. LeeDepartment of Electrical and Computer Engineering, San Diego State University, San Diego, California92182

(Received 8 M ly 1985; accepted for publication 15 July 1985)

Electrons proL.uced by two-photon ionization of CS,, SO2 , and (CH3)3N in N2 and CH4 buffer

gases at 193 nm were investigated using a parallel-plate drift-tube apparatus. At a low charge

density, the transient voltage induced by electron motion between the electrodes is proportional tothe gas pressure and the square of laser power. The two-photon-ionization coefficients measuredfrom the number of electrons produced are 3.3 X 10 - 27, 8.3 X 10- 30, and 1.7 X 10 - 27 cm 4/'W forCS2, S02, and (CH3)3N, respectively. The coefficient for (CH3)3N agrees with the earlier value

measured using ion current. At a high charge density, the number of electrons observed deviatesfrom the square dependence of laser power. The numbers of ions and electrons are greatly reducedby charge recombination whose reaction rate is enhanced in the presence of space charge.

INDENT *- !NDENT

I. INTRODUCTION II. EXPERIMENTAL

When molecules are irradiated by intense ultraviolet The experimental setup has been described in a previous

(UV) laser photons, high concentrations of charges can be paper.2 In brief, excimer laser (Lumonics Model 861S) at 193

produced. This charge production method has been used in nm was used to photoionize molecules in a gas cell of 6-in.

several applications, for example: (i) in the development of six-way aluminum cross. Only the central portion of the la-laser-triggered discharge switches that have low jitters (I ser cross-sectional area perpendicular to the laser beam was

ns)', (ii) in the measurements of electron attachment rate used for the experiment, whose size was 6.2 mm perpendicu-

constants of various electronegative gases in high-pressure lar to the applied field and 13.1 mm along the field. A rectan-buffer gases,2 and (iii) in the study of fundamental phenom- gular stop was used to define the beam area for which aena of electron leakage from a plasma.' In these applications, fraction of about 65% of the laser beam was used in the

the two-photon-ionization (TPI) coefficient is needed, be- experiment. The laser energy per pulse was measured right

cause it determines the absolute charge density. In this pa- after the exit window (suprasil) of the cell by an energy meterper, we report the TPI coefficients for molecules that have (Scientech Model 365). For the study of laser power depen-potential applications in electrical discharges. dence, the laser power was uniformly reduced by inserting

Although the multiphoton ionization (MPI) process has quartz plates in the laser path. The quartz plates were UVbeen extensively applied in the study of atomic4 and molecu- grade and polished so that they would presumably reducelar5 spectroscopy, there are relatively few measurements in the laser beam uniformly, namely, the beam profile was not

absolute MPI coefficients. In most earlier MPI measure- affected by the plates. Two parallel plate electrodes made of

ments - 9 positive ions were detected. In contrast, we mea- stainless steel of 5 cm diameter and 4.5 cm apart were used tosured the TPI coefficients here by detecting electrons. For a monitor the number of electrons. A negative high voltage

" comparison with earlier measurements,8 the TPI coefficient was applied to the cathode. The electron current induced byof trimethylamine [(CH 3)3N] was remeasured with this new the electron motion between the electrodes was converted

method. This method is extended to measure the TPI coeffi- into a transient voltage pulse across a resistor (R = 100-cients of CS2 and SO2. 1000 fl) connecting the anode to ground. The transient vol-

In the previous measurement' of trimethylamine, the tage was monitored by a storage oscilloscope (HP Model

ion current was limited by charge recombination when the 1727A) and photographed for permanent record.

"- charge density was high. This charge recombination process The gas pressure (250-300 Torr) in the gas cell waswas repeatedly studied in this experiment in a much shorter maintained constant in a slow flow system (flow rate about

time scale, because electrons drift much faster than ions. We 20 cm 3STP/min). The flow system reduces the buildup offind that the charge recombination rate is greatly affected by impurities released from walls or electrodes as well as those

space charge.3 The space charge can induce an internal elec- possibly produced from the photofragments of gases. The

tronic field to cancel out the applied field. The effective elec- laser repetition rate was usually low (typically 0.5 Hz), suchtric field in the plasma region is thus reduced such that the that the gas replacement rate was fast enough and the photo-electron energy decreases. Since the charge recombination fragments did not cause a serious problem. The gas pressure

rate is inversely proportional to the square root of electron was measured by an MKS Baratron manometer. All mea-energy,'o the reduction of electron energy by space charge surements were done at room temperature.

will increase the charge recombination rate greatly. This en- Sulfur dioxide (SO 2 ) or carbon disulfide (CS,) was dilut-

hanced charge recombination process is discussed in tH,; pa- ed in N, or CH4 before introducing into the gas cell filled

per. with the buffer gas of N, or CH 4 The N, and CH,4 gases

(supplied by MG Scientific) have minimum purities of CT

99.998% and 99.99%, respectively. The SO 2 supplied by N, (0) = ao j 2(j)dtnA1/hv, (1)Matheson has a minimum purity of 99.98%. The CS2 liquid(99.9% purity) was supplied by MCB Manufacturing Chem- where a (cm'/W) is the TPI coefficient, (W/cm2) is the laser

ists, Inc. The CS2 liquid contained in a stainless steel cylinder A (0.81 cm2 ) is the laser beam size, 1 (5 cm) is the laser pathwas pumped several hoirs at dry ice temperature to remove lg betwee the leres, I m is the p aer pathdissolved impurities bc fore its vapor was used for experi- length between the electrodes, hv is the photon energy (J),and T is the laser pulse duration.ment. The impurity level was believed to be very. low, be-cause the experimental results were reproducible in spite of The TPI coefficient can be determined from the electronthe measurements carried out before or after an additional number measured, if other quantities in Eq. (1) are known.pumping. The impurities that possibly accompanied CS2 va- The time profile of the laser pulse (supplied by the laser man-

por should not be more than 0.1%. A diluted trimethyla- ufacture) is used to determine fI dt, which is approximated

mine (0.05% in prepurified N2 supplied by Matheson) was by

used for the trimethylamine experiment. TI 2 (t)tt=XI At, (2)

II. RESULTS AND DISCUSSION where Ii is the laser flux at every At = 1 ns. The laser pulseA. CS2 width was about 10 ns, and the pulse profile was quite repro-

Electrons produced by laser ionization Of CS2 were ducible as specified by the manufacture. From shot to shot," monitored by the transient current induced by electron mo- the laser energy varied within 6%. Since only the central

tion between the electrodes. The transient current was mea- portion of the laser beam was used, the laser flux was as-sured by the voltage pulse across a resistor connecting the sumed to be spatially uniform. This makes theA value in Eq.anode to ground. The transient voltage pulses for CS,-N 2 (1) equal to the laser beam size in the central region betweenmixture at E/N = 9.6 Td (1 Td = 10-17 V cm2) are shown in the electrodes.

Fig. 1, where the laser energies were (a) 1.25, (b) 1.63, and (c) The number of electrons can be determined from the2.19 mJ/pulse in the region between the electrodes, and the transient voltage as,

pressures of CS, and N7 were 12 mTorr and 280 Torr, re- V(t) = i(t )Rf(t) , (3)spectively, where i(t) is the transient current induced by electron mo-

As shown in Fig. 1(a), the measured voltage V(t) is al- tion, R is the resistor connecting the anode to ground, andf(t)most constant before t = 1.5 ps, and after this time the cur- is the response of electronics which approaches unity when trent starts to drop because electrons hit the anode. This tran- is larger than the RC constant (C- 3 X 10 o F). The R val-sient pulse was produced by a low laser power, so the ues used in the experiment were in the range of 1010000 fl,electron concentration was low and the measured voltage corresponding to RCconstants of(3X r0g-3e 10- ) s.

was only slightly affected by space charge. [Figures 1(b) and The induced current depends on the number of elec-l(c) show the effect of space charge and will be discussedlater.] For the case of low charge density, the number of tosN()adteeeto rf eoiyWselectrons produced by each laser pulse through the TPI pro-cess can be described as8

200

TIME (,s)

.. 0 2 40>

E zo-- 100-

? >b

-20-

00 I 2

40RESISTANCE (1000 li)

FIG. 1. The electron conduction pulses of CS, at E/N =9.6 Td, where elec-* tron%%were prodtucedhs 1tAo-pholoniioiiization of 12.miTorr CS, in 280-Torr FIG. 2. The t%%o-photon-ion izat ion sifgnal vs external resistor R. The data

N, huffcr ga% The ArlJaer cnrgie% %erefaw 1.25, (b) 1 0'3 ,,nd (c) 2.1 3niJ/ were taken with laser encrr) of 3.3 mJ/pulse. E/- 7.1 Td. ICSl 5pulse vihe ektrode %pdsing was 4.5cmi alld the extert, csworwas V3OLI rnTorr, and (NJ 2W0 1 or.

• 9&

i(:t)=eN, (t)W/d, (4) 300

where d is the electrode spacing. In Eq. (4), we assume that '... 'all electrons are in equilibrium with the gas in the gas cell.

Substituting Eqs. (2H4) into (1), we have

V V=a(enAR W1',IIArY4(vd) , (5) 10

where Vis measured at a time sufficiently long after the laser E 5 0

pulse such thatf(t )-- I but before the electrons hit the anode

such that N, (t) = N, (0), namely, the Vvalues were measuredat the time shortly before electrons hit the anode.

In order to make certain that the ri-asured Vvalue wasnot affected by the response of electronics, we measured Vasa function of R. The results are shown in Fig. 2, which were o. ... .. .....measured with a gas mixture of 5 mTorr CS2 in 280 Torr N, 1 5 10 20

E/N = 7.1 Td, and a laser energy of 3 mJ/pulse. As shown LASER ENERGY (mJ/pulse}in Fig. 2, Vis linearly dependent on R except at R = 2X 10' 1Wn=NT

fl, where the measured Vvalue is slightly lower than expect- FIG. 4. The two-photon-ionization signal vs laser energy for CS2 in CH,,

ed. This result indicates that the measured V values at where[CSJ = 7.3mTorr,[CH4J = 25OTorr, E/N= 13.2Td. andR = 100

R < l0' fl are not affected by the response of electronics. The 11.dependences of Von other parameters (such as electron driftvelocity, laser flux, and gas pressure) are further checked the electron drift velocity shows on observable change onlybelow. when CS2 in N2 or CH4 is more than 0.5%. In this experi-

The electron drift velocities for various amounts of CS2 ment, the fraction of CS2 was always smaller than 0. 1%,or S02 in N, or CH4 were measured as a function of E /A in thus, the electron drift velocity of the gas mixture was notthis experiment, in which electrons were produced by irradi- different from the buffer gas.ating the cathode with a KrF (248 nm) laser photons. (At this The dependence of the observed voltage (which waswavelength, the photon energy is not enough to ionize CS2 measured at 1.5 s as shown in Fig. 1) on laser energy was

and o2 by the T process.) The electrons produced by the investigated. The results for CS2 (4.9 mTorr) in N2 (280 Torr)athode are well localized in space, so that the electron drift are shown in Fig. 3, where E IN = 7.1 Td and R = 1000 fl.

time is well defined. The electron drift velocity is equal to the Each datum point in Fig. 3 represents a single laser pulse. Atelectrode spacing divided by the electron drift time. Since the laser energies less than 3 mJ/pulse, V is linearly dependentconcentrations Of CS2 and SO2 used in the measurements on 1 2. This indicates that electrons are produced by the TPIwere so small that the electron drift velocities we measured process. At high laser energy, V deviates from the P2 depen-for various gas mixtures were essentially the same as that of dence. The Vvalues decrease greatly at laser energies higherpure N,(Ref.12) or CH4(Ref. 13) gas. More quantitatively, than 5 mJ/pulse. The decrease is caused by charge recom-

bination and change of electron drift velocity which are en-hanced by the space-charge effect (see discussion later). Asnoted before, the pulses shown in Figs. I(b) and I(c) have two

.. _ maxima. We measure the V value at the later maximum,because it was less affected by the response of electronics andthe space-charge effect.

Similarly, the dependence of voltage on laser energy wasalso investigated for the gas mixture of CS2 (7.3 mTorr) inCH, (250 Torr) as shown in Fig. 4, whereE/N = 13.2 Td and

.,R = 100 fl. Again, at low laser energy, Vis proportional to.,' 12, and at high laser energy, V decreases with increasing laser

E energy. This decrease is again caused by charge recombina-

> tion and change of electron drift velocity.It should be noted that there was a background signal

before CS2 was added to the buffer gas in the gas cell. Thebackground could be from the ionization of residual gas in

0.1 0.5 1 5 1o the cell and /or from irradiation of the electrodes by straylaser light. The background signals were about the same for

LASER ENERGY ( mJ/pulse) both N2 and CH, buffer gases. The background signal wasabout 5% of the TPI signal of CS2, and was subtracted out

• FIG. 3. The two-photon-ionization signal vs laser energy for CS, in N, from the total signal. Thus the data shown in Figs. 3 and 4where ICS:] - 4.9 mTorr, IN] 28O lorr, E/N 7.1 Td, and R 1000 are free from the background signal. It should also be noted

that the laser energies shown in Figs. 3 and 4 are the true

S.7.': UT ' 4' A IiA.,"

is a two-photon process, and Eq. (5) is applicable for the case160 of low charge density. Thus, we can measure the TPI coeffi-_60- b cient using the data taken at low laser energy and low CS2

pressure. It should be noted that the loss of electrons by

attachment to CS2 at low pressure is negligible. This is justi-• fled by the fact that the transient pulse shown in Fig. 1(a) is

I flat in the 0-1.5-pis region. If the attachment process occurs,E 80 the transient voltage will decrease monotonically with time.2

>This assertion is consistent with the fact that the mean elec-tron energies ( < 1 eV) for N2 and CH4 at EIN<20 Td aremuch lower than the dissociative electron attachmentthreshold for CS2 (3.09 eV). 1

The TPI coefficients measured with the CS2-N2 and0 8 e CS2-CH gas mixtures are 3.09 X 10-and 3.1 X10- 27 cm4/0 8 16W, respectively. These values are determined from the linear

C S2 ] (m t or r) slopes shown in Figs. 3 and 4, corrected with the preabsorp-tion of laser energy by CS 2 in Fig. 5. These coefficients along

FIG. 5. The two-photon-ionization signal vs CS2 gas pressure, where with the ionization potential'8 and the single-photon absorp-[CH.] = 250 Torr, E/N = 13.2 Td, and R = 100 fl. (a), the laser energy in tion cross section of CS2 at 193 nm measured in this experi-the central region between the electrodes is attenuated by CS2; (b) the signalsin (a) are corrected such that the laser energies are all equal to 3.1 mJ/pulse ment are listed in Table I. The TPI coefficients measuredin the central region between the electrodes, with various laser energies and gas pressures vary within the

experimental uncertainty which is estimated to be + 20% of

laser energies at the central region between the electrodes, the average a value of 3.3 X 10-27 cm 4/W. The sources ofThe window transmission (about 90%) and the absorption of major experimental errors are (i) gas concentration (5%), (ii)CS2 have been corrected. The photoabsorption cross section electron drift velocify (10%), (iii) laser flux (5%), and (iv)ofCS2 at 193 nm measured here is L.OX 10-6 cm2 , which is background signal (1%).smaller than the value given by Rabalais et.al.'4 by a factor of The ionization potential ofCS2 is 10.08 eV. " The energyabout 2, but our value is consistent with the measurement of of two ArF laser (6.42 eV/photon) photons is only sufficientSuto and Lee's using synchrotron radiation as a light source. to ionize CS, into CS, +, but not energetically possible to

The dependence of the TPI signal on CS2 pressure (in N 2 produce other fragment ions such as CS+, S', and S2 '. Thisor CH4 buffer gas) was investigated. As an example, the re- is different from the observation of Seaver et al.'9 that CS +

suit of CS, in 250TorrofCH4 isshown in Fig. 5(a),whereE/ was a major ion and significant amounts of S-, S, + andN = 13.2 Td, R = 100 fl, and the laser energy was kept con- CS2 ' were produced from the multiphoton ionization ofCS2

stant at 3.1 mJ/pulse. Due to the attenuation of laser intensi- at 193 nm and 266 nm. In their experiments, the lasers werety by CS2, the laser energy in the central region of the gas cell focused and their laser flux of 101° W/cm2 was about fourdecreases with increasing [CS2]. Thus, the observed voltage orders of magnitude higher than the flux used in this experi-is not linearly dependent on [CS21]. However, if the photoab- ment. With such a high laser power, fragment ions can besorption of CS2 is corrected such that the laser energies are produced by the stepwise photoabsorption processes. Inthe same, then the corrected Vvalues become linearly depen- each step the absorption can be saturated (i.e., all moleculesdent on [CS2 ] as shown in Fig. 5(b). Thus, our results are are excited) such that the power dependence deviates fromconsistent with the expectation of Eq. (5) that V increases I ", where n is the number of photons that are energeticallylinearly with n. possible to produce the observed fragment ion. For example,

From the above studies for the dependences of the ob- it requires three ArF laser photons to produce CS', but theserved voltages on the response of electronics, the laser flux, laser power dependence 9 is only n = 1.1. Since our laserand the gas pressure, it is conclusive that the photoionization intensity is relatively low and the laser power dependence of

TABLE !. Two-photon-ionization coefficients a for CS., SO., and trimethylamine at 193 nm. Ionization potentials 1. P. and single-photon absorption cross-sections a at 193 nm are also listed.

a(cm4'/WiI.P. a

Molecules (eV) (cm2; in N. in CH. Ref. 8

CS, 10.08, l .OX 10- 3.4x 10- 3.1 x 10-"SO., 1 2 .3 1h 9.0X I0-" 8.6x l0- Ox 10-

(Clt,(,N 8.5' 1.04 X 10 -7d 1.7x 10-2' 1.5> 10-"

'Reference 18.'Reference 22.

'Rcferenc 24SRefe'rence

d~c~tr~nI 'M_

------------- -

30 tages increase with the square of laser energy. This result .again shows that electrons are produced by the TPI process.In contrast to CS2, there is no space-charge effect occurring

at higbhlaser energy, because the TPI coefficient of S02 is sosmalf that the charge density is always low.

107 The dependence of the TPI signal on the SO2 pressure isb -. *shown in Fig. 7(a), where [N2] = 250 Torr, E/N = 10.8 Td,.-,0 laser energy = 7.21 mJ/pulse, and R = 1000 f. Similar to

E .- the case of CS,, the TPI signals do not increase linearly with>/ "increasing [SO2J, because the laser energy is preabsorbed by

SO 2. The absorption cross section measured at 193 nm is9.0X 10- 18 cm 2, which is comparable with the publishedvalue.20 If the laser energy is corrected, the voltages pro-duced by a same laser energy of 7.21 mJ/pulse are shown inFig. 7(b). The voltages increase linearly with increasing5 10 [S02] as expected from Eq. (5). The electron attachment toLO2 is negligible, because the electron energy in N2 or CH 4

(mJ/puse) ( < I eV) " is too low to make the electron attachment occur.

The threshold energy for the dissociative electron attach-FIG. 6. The two-photon-ionization signal vs laser energy for SO 2 in N 2 and ment process of SO2 is about 3.5 eV.2"CH.. In (a), [SO 2) = 33.6 mTorr, [N2] = 250 Torr, E/N= 10.8 Td, and The TPI coefficients measured with the S0 2-N2 andR = 1000fl. In (b), [S0J = 32 mTorr, JCH4 ] = 25OTorr, E/N= 10.8 Td, SO2-CH3 mixtures are 8.6x 1030 and 8.0x 10-30 cm4/W,and R =330 [1. S2Cmxue r .X ( n .X

respectively. These values along with ionization poten-electrons produced is 12,CS2 + is believed to be the major ion tial22 23 and absorption cross section are listed in Table I. Theeleon s xperoued. ivalues measured at various laser energies and gas pressuresin our experiment.

vary within experimental uncertainty which is estimated to -We have tested the three-photon-ionization process us- be ± 20% of the average a value of 8.3 X10-30 cm'IW.

ing KrF laser photons at 248 nm (5.0 eV). The transient vol- This coefficient is more than two orders-of-magnitudetage signal was too small to be detectable. The upper limit of smaller than that of CS2. The experimental error caused bythe three-photon-ionization coefficient of CS2 at 248 nm was the background signal is higher than that of CS2. However,estimated to be 7 X 10- 9 cm 6/V 2. the overall experimental uncertainties are not much differ-

B. SO ent for both molecules, because the experimental error forthe background signal is small when compared with other

The transient voltages produced by two-photon ioniza- errors.tion of SO by various ArF laser energies are shown in Fig. 6. 1The data were taken with [SO 2] = 33.6 mTorr, [N2] = 250 C. (CH3 )3NTorr, E/N= 10.8 Td, and R = 1000 11 for Fig. 6(a), andwith[SO2 =32mTorr,[CH] = 250Torr, E/N= 10.8Td, In order to compare with the earlier measurement, theand R = 330 fl for Fig 6(b). In both cases, the transient vol- TPI coefficient of trimethylamine was also measured in this

experiment. For low laser energy, the voltages increase withthe square of laser energy as expected. The dependence of the

sC

bBb A

A

40 I >E 4-

0 100 200 o

[SO2 ] mtorr)0 0 (mor 20[TRIMETHYLAMINE] (mtorr)

FIG.7. The two-photon-ionization signal vs SO2 gas pressure, where(N:] = 250 Tort, E/IN = 10.9 Td, and R = 1000 (1. The signals in (b) arc FIG. B. The two-pholon-ionization siri'al v% trmethylamine Pas Presu C.

corrected from fal such that tie lar energics at the central region between where [N: ] -2N soirr, /A 7.06 1 d. R - I(j 11. arid Il ie lascr c i'r rythe elcctrods arc all equal to 7.21 rri/pulse. of 2.06 nrn/puk1(.

" "r'k' " " ' ' ' ' E ,,,;~7 " " .,"

' " . ..

transient voltage on the trimethylamine pressure is shown in TIME (Ps)Fig. 8. The trimethylamine pressures are so low that thepreabsorption of laser enery is negligible and the voltagesincrease linearly with gas pressure. 0 - \ 4 .. 1:: _

The measured a value is 1.7)X 10-27' cni'/W as listed in I >

Table I where the ionization potential2' and the absorption .5i " . I',

cross section' are also li ited. The current a value agrees very > Awell with the earlier measurement' of 1.5X 10-2 cm4/W,despite that these two measurements were carried out bydifferent methods. In the earlier experiment' the ion current -200was measured, and in this experiment we measured the num-ber of electrons. The good agreement indicates that the pres-ent method is reasonably reliable. FIG. 9. The transient voltage pulses produced by the two-photon-ioniza-

It is of interest to note that at 193 nm,CS2 has a TPI tionofCS2 inCH4., where[CS21] =7.3nTorr, E/N= 13.1 Td, R = 100(1,

coefficient higher than trimethylamine by a factor of 2. and [CH.] = 257 Torr. The laser energies were Ia) 2.6, (b) 5.5, and (c) 8.0i/pulse.

Thus, similar to trimethylamine, CS, is also a good candi-

date for producing high-electron density in laser-controlled The current reverse has been observed in an earlier tri-switches. methylamine experiment.3 This current reverse is likely

caused by the space-charge effect, which will be described inD. Charge recombination detail in another paper. Here we present a brief scheme to

It was observed' that at high laser energy the ion current explain the observed phenomena. The first peak in Fig. 9(c)produced by two-photon ionization of trimethylamine was was produced by the electrons initially produced by lasersmaller than the value calculated from the 12 dependence (see ionization and drifted in the applied field. As soon as elec-Fig. 3 in Ref. 8). The decrease of ion current was attributed" trons move, the electrons and the ions in the opposite plasmato charge recombination. This process was further studied in fronts form a quasidipole that produce an internal electrica short time scale by monitoring the number of electrons field in the direction opposite to the applied field. The totalbetween the electrodes.3 It was found that the electrons and electric field in the plasma region, which is equal to the sum

ions in the plasma fronts initially drifted apart by the applied of the applied field and the space-charge-induced field, isfield could form a quasidipole to induce an internal electric thus reduced so that the electron drift velocity decreases.field opposite the applied field. Depending on the charge This explains the voltage decrease after the first peak.density, the induced field could be large enough to suppress The space-charge-induced electric field depends on thethe applied field. Once the total electric field in the plasma charge density. If the charge density is high enough, the in-region is reduced, the electron-ion recombination rate induced field at some instant can be so large that it over sup-creases. This space-charge effect gives a reasonable explana- presses the applied field. Namely, the total electric field cantion for the decreases of ions and electrons observed in the be in the direction opposite to the applied field such that theexperiments of trimethylamine.3." majority of electrons (which include the electrons both in-

The space-charge-induced charge recombination pro- side and outside the plasma region) move toward the cath-cess was again observed in the CS2-N2 gas mixture. Figure ode. This explains the change of voltage sign from negative1(a) shows a slight decrease in the central portion of the to positive (current reverse) as shown in Fig. 9(c). As timepulse. This minimum becomes more obvious in Figs. 1(b) and proceeds, the charges in the plasma region may be lost byl(c) when the laser energy increases from 1.25 mJ/pulse in charge recombination and the space-charge-induced fieldFig. l(a) to 1.63 mJ/pulse in Fig. 1(b), and to 2.19 mJ/pulse may be weakened by a large separation of charges, thus thein Fig. 1 (c). When the laser energy further increases, the min- majority of electrons move toward the anode again. This

. imum can even change sign to positive. explains the appearance of the second peak. After those elec-'. Similar results were observed in the CS2-CH4 gas mix- trons originally in the plasma front reach the anode, the cur-

tures as shown in Fig. 9, where [CS2] = 7.2 mTorr, rent is dominated by the electrons in the plasma region that

[CH4] = 257 Torr, and laser energies were 2.6, 5.5, and 8.0 still move toward the cathode. This causes the change ofmJ/pulse for Figs. 9(a), 9(b), and 9(c), respectively. The vol- voltage sign again from negative to positive (current reverse)tages of the first peak at a short time after the laser pulse after the second peak as shown in Figs. 9(b) and 9(c). (Theseincrease with laser energy. But the second peak first in- results indicate that electrons carry a sufficient inertia mo-creases with laser energy, and then decreases at high laser mentum such that they do not respond instantaneously withenergy. (The laser energy dependences of the voltage for the the change of electric field.) The long decay time of the posi-second peak are shown in Figs. 3 and 4.)The minimum in the tive voltages as shown in Figs. 9(b) and 9(c) suggests that thecentral portion of each pulse becomes deeper as the laser loss of electrons is due to the charge recombination process.energy is sufficiently large. The voltage even changes sign Although this scheme explains the observed current reversewhen the laser energy is large enough as shown in Fig. 9(c). qualitatively well, quantitative analysis to support this claimThe voltage (or current) reverse means that the majority of is necessary and will be studied further.electrons mo~e toward the cathode instead of moving to- In the measurement of the TPI coefficients, the space-ward the anode by the applied field. charge effect was kept as low as possiblc. This was accom-

............. -. ;.......--. ....-..--........................................ ~~~~~~ J7 ( -* - -

- plished by keeping the charge density low or increasing the 'L. C. Lee and F. Li, "Space Charge Effect on Electron Conduction Cur-

applied electric field. The linear portions shown in Figs. 3, 4, rent" (unpublishedj.'G. S. Hurst, M. G. Payne, S. D. Kramer, and J. P. Young, Rev. Mod.

and 6 obey the TPI process of square-laser-power depen- Phys. 51, 767 (1979); C. H. Chen, G. S. Hurst, and M. G. Payne, Chem.

space-charge effect. Thf. TPI coefficients measured under 'P. M. Johnson, Appl. Opt. 19, 392011980); P. M. Johnson and C. E. Otis,

such conditions shoulc be independent of E/N applied, Ann. Rev. Phys. Chem. 32, 139(1981).',. H. Brophy and C. T. Rettner, Chem. Phys. Lett. 67. 351 (1979).

which were confirmed in our experiments. 'W. Zapka and F. P. Schafer. Appl. Phys. 20. 287 J1979).*. a' j. C. Lee and W. K. Bischel, J. Appl. Phys. 53, 203 (1982).

IV. CONCLUDING REMARKS 'G. A. Abakumov, B. 1. Polyakov, A. P. Simonov, L. S. Tchuiko, and V. T.

Yaroslavtzev, Appl. Phys. B 27, 57 (1982).The two-photon-ionization coefficients of CS2, SO 2, and 10J. A. Rees, in Electrical Breakdown of Gases, edited by J. M. Meek and J.

(CH,)3 N have been measured by observing the transient cur- D. Craggs (Wiley. Chichester, UK, 1978), p. 70-76.rent induced by electron motion in high-pressure buffer gas 1L. G. H. Huxley and R. W. Crompton, The Diffusion and Drift of Elec-

number otrons in Gases (Wiley, New York, 1974)(N 2 or CH . The dependence of the nu rof electrons on 12. Dutton, J. Phys. Chem. Ref. Data 4, 577 (1975).laser energy and gas concentration was investigated. At low "w. C. Wang and L. C. Lee, J. Appl. Phys. 57,4360 (1985).

charge density, the transient voltage is proportional to gas ",J. W. Rabalaib, J. M. McDonald, V. Scherr. and S. P. McGlynn, Chem.

pressure and the square of laser energy. This shows that the Rev. 71, 73 (1971).cpM. Suto and L. C. Lee (unpublished).

charges are produced by a two-photon-ionization process 1L. G. Christophorou. Atomic and Molecular Radiation Physics (Wiley,

whose coefficients are reported in this paper. At high charge New York, 1971).

density, the transient voltage is reduced by charge recombin- "J. P. Ziesel, G. J. Schulz, and 3. Milhaud, J. Chem. Phys. 62, 1936 (1975).

ation whose reaction rate is enhanced in the presence of 'W. C. Price and D. M. Simpson, Proc. R. Soc. 165,272 (1938); Y. Tanaka.A. S. Jursa, and F. J. LeBlanc. J. Chem. Phys. 32, 1205 (1960).

space charge. 19M. Seaver, J. W. Hudgens, and J. J. Der- rpo. Chem. Phys. 70,63 (982);ACKNOWLEDGMENTS Int. J. Mass Spectrom. Ion Phys. 34, 159 (1980).

"GD. Golomb, K. Watanabe, and F. F. Marmo, J. Chem. Phys. 36, 958

The authors wish to thank Dr. E. R. Manzanares, Dr. J. (1962); D. E. Freeman, K. Yoshino, J. R. Esmond, and W. H. Parkinson,B. Nee, and Dr. M. Suto in our laboratory for useful sugges- Planet. Space Sci. 32, 1125 (1984).

2j. Rademacher, L. G. Christophorou, and R. P. Blaunstein, J. Chem. Soc.

tions and discussion. This work is supported by the Air Faraday Trans. 11 71, 1212 (19751.

Force Office of Scientific Research under Grant No. "J. H. D. Eland and C. J. Danby, Int. J. Mass Spectrom. Ion Phys. 1, 111

AFOSR-82-0314. (1968); D. W. Turner, C. Baker. A. D. Baker, and C. R. Brundle, Molecu-lar Photoelectron Spectroscopy (Wiley, London. 1970;, pp. 84-86.

'J. R. Woodworth, C. A. Frost, and T. A. Green, J. Appl. Phys. 53, 4734 230. J. Orient and S. K. Srivastava, J. Chem. Phys. 80, 140 (194).(1982). M. B. Robin. Highly Excited States of Polyatomic Molecules, Vol. l and 11

2L. C. Lee and F. Li, J. Appl. Phys. 56, 3169 (1984). (Academic, New York, 1974).

FOOTr 'OTELI;NE L IM IT _._...

iS .

Appendix B

"Electron Attachment to H20 in Ar, N2 andCH4 in Electric Field"

umrjUvw U L

Electron attachment to H20 in Ar, N2, and CH 4 in electric fieldW. C. Wang and L. C. LeeDepartment of Electrical and Computer Engineering, San Diego State University, San Diego, California92182

(Received 25 October 1984; accepted for publication 19 December 1984)

The attachment of electrons to H20 in Ar, N2 or CH4 is investigated using a parallel-plate drift-tube apparatus. Electrons are produced either by irradiation of the cathode with ArF laserphotons or by two-photon-ionization of a trace of trimethylamine in a buffer gas. The transientvoltage pulses induced by the electron motion between the electrodes are observed. The electronattachment rate of H20 is determined from the ratio of transient voltage with and without H20added to the buffer gas. The measured electron attachment rate constants of H2 0 in Ar increasewith E/N from 2 to 15 Td. Electron attachment due to the formation of "temporary" negativeions in the H20-N 2 and H20-CH4 mixture were observed. The lifetime of the negative ion wasdetermined to be about 200 ns, whose nature is discussed. The "apparent" electron attachmentrate constants for the formation of "temporary" negative ions in the H20-CH4 gas mixture aremeasured for E IN from 1 to 20 Td. The electron drift velocities for the gas mixtures of H20 invarious buffer gases are measured.

I. INTRODUCTION evident by electrons being attached by H 2 0 in N 2 or CH4 at

Recent advances in pulse-power technology indicate' low E IN.

that opening switches are needed for further developing a In this work, we applied a relatively new method' 9 to

magnetic energy storage system. The inductive energy investigate the electron attachment in mixtures of water

storage is preferable to the capacitive storage because the vapor in Ar, N2, and CH4. The mean electron energies inenergy density in the inductive system is much higher the buffer gases decrease in the order of Ar, N2, and CH4 at(some 10-10s tn t he in citive system .2 Opening each fixed E IN. For example, the mean electron energies at

(some~~~~~~~~~~~~~ I0~0 ie)ta h aaiiesse.Oeig EN = 1.2 Td are 2.53, 0.35, and 0. 11 eV for Ar, N2, andswitch requires a fast decay of conduction current which CH, respetive 2 Thus, we were e to Ar, the and.' ', oul beacheve byattaching electrons to electronegative CH4, respectively.' 2 Thus, we were able to study the elec-

*could be achieved by atahn lcrn oeetoeaie tron attachment process of water vapor over a wide rangegases mixed in a buffer gas.' Thus, the electron attachment tr n en r es °f w te V).rates for various gas mixtures are needed for the develop- of electron energies (0.1-10 eV).ment of this type of discharge switch. Our study of the 11. EXPERIMENTelectron attachment due to a small amount of H 20 in Ar,

. . N 2, and CH4 is reported in this paper. In this experiment, electrons were produced by irradia-The dissociative electron attachment process of water tion of the cathode with ArF laser photons which has a

vapor has been studied extensively by the swarm meth- pulse duration of about 10 ns. The laser pulses are quiteod3-0 and the electron-beam method." The predominant uniform in both the spatial and temporal distribution ofnegative charged species are H- and 0- ions with the on- photon fluxes. For the study of the H20-N 2 mixture, elec-set energies of about 5.5 and 4.9 eV, respectively.' 2" 3 Be- trons were also produced by two-photon ionization (TPI) ofcause these onset energies are high, the density-reduced a trace of trimethylamine present in the buffer gas usingelectric field, E IN, in pure H20 must be larger than 40 Td ArF laser photons. The experimental set-up has been dis-(lTd = 10- 17 Vcm 2 ) in order to have electron attachment cussed in detail in a previous paper."9 Briefly, the gas cell isoccur. The E IN can be lower if Ar buffer gas is used, be- a 6-in. six-way aluminum cross. The electrodes are twocause for same E/N the electron energy 2 in Ar is higher parallel stainless steet plates of 5 cm in diameter and 3.7 cmthan in H20. Earlier studies on the electron attachment apart. The energy of ArF laser pulse (Lumonics Modelhave been summarized by Gallagher et al.'' Herein we re- 86 IS) was monitored by an energy meter (Scientech Model

" port additional measurements on electron attachment 365). A negative high voltage was applied to the cathode.- rates for the dissociative attachment process of H20 in Ar The conduction current induced by the electron motion

at varied electric fields. between the electrodes was observed as a transient voltageBradbury and Tatel 5 observed an electron attachment pulse across a resistor (220-1000 D) connecting the anode

in water vapor at low electron energy (less than I eV). The to ground. The transient pulse was monitored by a 275attachment probability increases with the HO concentra- MHz storage oscilloscope (HP Model 1727A). Each singletion, [H20], and decreases with increasing E/N. A similar transient pulse was stored in the oscilloscope which wasresult was later reported by Kuffel.3 The attachment is at- later photographed for a permanent record. This experi-tributed to nondissociative process that leads to the forma- mental method is similar to that used by Verhaart and Vantion of negative ion. However, no long-lived HO- ions der Laan,2"' Christophoru et al.,2' and Aschwanden,12 forwere observed by several other works."- • 12 "This discrep- the electron ionization and electron attachment measure-ancy could be reconciled if there exits a short-lived negative ments.ion, as suggested by Hurst et al. '" From our observations The gas pressure in our gas cell was maintained constantwe conclude that this short-lived negative ion does exist, as in a slow flow system (flow rate -20 cm3 STP/min). The

4360 J. AppI. PhyS. 57 (9). 1 May 1985 0021-8979/85/094360-08$02.40 @ 1985 American Institute of Physics 4360

flow system reduces the buildup of impurities released E IN, electrons can reach an equilibrium state in a very shortfrom walls and electrodes as well as those produced from time. Thus, Wand W' can be practically considered as con-the photofragments of gases. The gas pressure (in the range stant except for those electrons near the electrodes. (At an Arof 200-500 Torr) was measured by an MKS Baratron ma- pressure of 400 Torr and an electric field of 100 V/cm, thenometer. All measurements were done at room tempera- relaxation time23 for reaching an equilibrium is about 10 ns.)ture. For t < d IW (or d /W'), N, (or Ae') is a constant except for

The Ar, N2, and CH 4 gases (supplied by MG Scientific) the beginning of the pulse where some electrons diffuse backhave minimum purities of 99.998%, 99.998%, and to the cathode.23 Therefore, after a reasonable time, the first99.99%, respectively. To produce electrons by the TPI term of Eq. (4) is independent of time, that is, the V(t) ismethod, a diluted trimethylamine (0.05%) in prepurified nearly constant. The plot of ln(V'/V) vs t thus gives the elec-nitrogen (supplied by Matheson) was used. All these gases tron attachment rate Vo.were admitted to the gas cell without further purification. In the absence of water vapor, no attachment occurs.The water vapor was carried by Ar, N2, or CH4 into the gas However, even if a small attachment due to impurities incell. The concentration of H20 was determined from the the buffer gas does occur, the attachment can be includedratio of water vapor pressure (25 Torr at 26* C) to the pres- in the background and does not show in Eq. (4), namely, thesure of carrier gases (> 1 atm). In order to remove the oxy- measured v value is only due to the gas added (H2 0 in thisgen dissolved in the water, the water in a container was case).pumped periodically for several days before the water va- When the electrons are produced from the laser irradia-por was used for the experiment. The partial pressure of tion of cathode, the electron drift velocity is equal to d IT,H20 used in the experiment ( < 5 Torr) was smaller than the where T is the electron drift time between the electrodeswater vapor pressure at room temperature. The carrier gas which can be determined from the transient voltage wave-bubbled through the water, and the flow rate was kept low form. In this experiment the mean electron drift time (0.4-4so that the equilibrium between the gas and the liquid of us) is much longer than the duration of laser pulse (-10 ns),water exists in the water container. To ensure that the H20 so the electrons are essentially all produced at t = 0. Theconcentrations used to determine the rate constants are effects of longitudinal and transverse electron diffusion oncorrect, we have also used a premixed gas mixture of H20 the measurements are neglected, since the gas pressures used(271 ppm) in Ar (supplied by Matheson) for the experiment, in this experiment are quite high. For example, the NDLThe attachment rate constants measured from both the value 2 1

26 for Ar in the present E/N region is aboutpremixed gas mixture and our water container are the 1.5 X 1022 cm- s-'. For an Ar pressure of 200 Torr and ansame. electron drift time of 1 ,as, the electron diffusion width

(4-W t ) is only about 0.07 cm. This is much smaller than

III. ANALYSIS METHOD the separation of electrodes (d = 3.7 cm). Therefore, the dif-

The method for the analysis of the data has been de- fusion effect is not significant and is negligible in the data

scribed in the previous paper.'9 In brief, the current 23 in- analysis. In any case, the diffusion effect does not affect theduced by the electron motion is expressed as measurement of electron attachment rate constants, al-

though it may slightly affect the electron drift velocity mea-

i(t) = eN. W/d, (1) surements.

where N, is the number of electrons between the elec- If trimethylamine is used to produce electrons 2' as introdes, W is the electron drift velocity, and d is the elec- the study of the H20-N 2 mixture, equal amounts of elec-trode spacing. trons and ions are produced by the TPI process in space.

When water vapor is added to the gas cell, electron at- Since ions move much slower than electrons, ions can betachment by water molecules occurs, and current becomes considered as stationary in the time scale of interest, and Eq.

(4) will thus be applicable in this case too. Nevertheless, the't= eN W 'e - 'Id, (2) charge density should be kept low to avoid the space charge

where v, is the electron attachment rate of water vapor, effect due to the charge-induced field.2" For this case, theSince the absorption coefficient of H2 0 at 193 nm (ArF electron conduction current is very sensitive to the laserlaser wavelength) is too small to attenuate the laser intensi- power fluctuation, because N, is proportional to 2, where Ity, N; is essentially equal to N, if the laser power is con- is the laser intensity. The concentration of trimethylaminestant. used in the experiment is usually small (< I mtorr), so its

The current is converted to transient voltage by electron attachment rate is negligible (actually, the attach-

Vt ) =f(t )i(t)R, (3) ment due to trimethylamine is included in the background

where R is the resistor connecting the anode to the ground signal).

and f(i ) is the response function of the detection system thatapproaches constant as t > RC (- 60 ns). The logarithm of IV. RESULTSthe ratio of the voltages with and without water vapor is A. H20-Ar mixture

• i thustuThe electron transient waveform V(t) for the electron

ln(V'/V) = In(N; W'/N. W) - vat. (4) motion in 200 Torr of Ar atE/N = 14.5 Td is shown in Fig.

For an atmospheric pressure and a relative low value of 1(a). The pulse is a single shot, but it is a representative for

4361 J. Appl. Phys., Vol. 57, No. 9, 1 May 1985 W. C. Wang and L. C. Lee 4361

, , ,- - :,-, .. / ; .? .v, -, . ,- ;- : .?k k . : ' " ; ': , " P 'C' " " a q '- i

" ~w- * i '

"' 'J''' - :" '' t ' ' , , = , ' = t' . . .

the means ofseveral subsequent shots. From shot to shot, the "pulse amplitudes vary within 10%. This variation is mainly 1.5-

caused by the fluctuation of laser energy. Nevertheless, the 110 U

pulse shape is very reproducible, which meets the require- C. ,

ment for our measurements. >" 1.0 0The voltage shown in Fig. (a) decreases rapidly after the " Ago

first peak, which is due to the loss of electrons by back > ,a,$

diffusion to the cathode.23 The voltage approaches a nearly r 0 . 1 .*&*Gconstant value as electrons move away from the cathode. C 0 VThe V(t) starts to drop when electrons arrive at the anode. 0The time that V(t ) starts to drop is used to measure the drift 0

time T. The amplitude shows a slight increase before it 0drops. This may be caused by the electrode effect, namely, 0 4 8 12 16

electrons are accelerated more by the image charges when E/ N ( 1017 V-cm 2 Ithey move close to the anode. Since the gas pressure is high FIG. 2. Theelectron drift velocities as a function of E/N for thegases with-

(200-400 Torr), the effect of electron diffusion is not large out (vI and with 0.15% of H 20 (0) in Ar, The electron drift velocity in-

as stated before, and it may not affect the T value more than creases when H20 is added to Ar. The data given by Errett Q and Nielsona0%. ste etron, adit vely in bet the palue Aor ghas (A) forpureAr and Hurst etaL. for [H20]/[Arl = 0(-), 0.137% (- - -),and10%. The electron drift velocity in both the pure Ar gas 0.25% .- ) are plotted for comparison.and the H 20-Ar mixture as a function of E IN are plottedin Fig. 2. For pure Ar, our measured drift velocities are The flat portion (1 < t < 3 .3 jus) in Fig. I(a) indicates thatsmaller than the published data,25 "29 30 which are also plot- the number of electrons drifting in space are constant andted in Fig. 2 for comparison. The data of Hurst et al.5 for their drift velocities reach an equilibrium, namely, N, Wthe Ar + H20 mixtures are also plotted in Fig. 2, which are = const. The decrease of the voltage in the flat portion dueconsistent with our measurements. Our data may be affect- to the addition of H 2 0 can thus be used for the measuremented by the large electrical noise created by the laser dis- of electron attachment rate. During the period of the flatcharge. The uncertainty for the t = 0 point is about 20 ns portion, the first term in Eq. J4) is independent of time andthat may introduce an uncertainty of about 5% for the theplotofln[V'(t)/V(t)]vstgivesv0 .TheratioofV'(t)/V(t)measured electron drift velocity, at EIN = 1.45 Td for with and without 0.024, 0.048, and

When 0.073 Torr of H20 is admitted to Ar at the same 0.073 Torr of H 20 in Ar are plotted in Figs. 3(a), 3(b), and

experimental condition of Fig. l(a), the pulse duration and 3(cl, respectively. It is worth noting that the extrapolated

the amplitude decrease as shown in Fig. I(b). The shortening value of V7 Vat t = 0 increase with increasing H2O concen-of pulse duration indicates the increase of electron drift ve- trations. This is caused by the increase of electron drift veloc-

locity and the decrease of amplitude is caused by the electron ity when H20 is added to Ar as stated before. As shown inattachment to water vapor. (The electron drift velocities Fig. 3, ln(V'/V) is linearly dependent on t, and the electronwhen adding H 2 0 to Ar at varied E/N are shown in Fig. 2.) attachment rate v. is obtained from the slope.The first peak in Fig. 1(b) is higher than that in Fig. l(a), The results for v. /[H 2O]asa function of[H20]/[Arj are

which again indicates the increase of electron drift velocity shown in Fig. 4. At a fixed E IN, the v. /[H 20] values de-when H20 is added to Ar. In fact, the amplitudes of the first crease with increasing [H 20]/[Ar], which is likely caused5

peak at various water vapor concentrations are found to be by the effect of H 2 0 on the electron energy distribution inlinearly dependent on the electron drift velocity as expected Ar. That is, an increase in [H20]/[Ar] will decrease thefrom Eq. 14) for t0. number of electrons in the energy range where the dissocia-

tive attachment process will take place. The attachment

Time (ps)

0 2

; 30E

01 2 3

FIG. I. The transient voltage waveforms produced from the electron mo- Time (ps)

tion in 200 Torr of Ar (al without H.0 and (bi with 0.073 Torr of H,O. The FIG. 3. The ratios of the transient voltages measured with and without H,O

electrons were produced from irradiation of the cathode by ArF laser pho- in 200 Torr of Ar, V'/V, as a function of the clasped time after laser pulse.tons. The E/N was fixed at 14.5 Td, the electrode spacing was 3.7 cm, and The (H,O1 are: (a) 0.024 Torr. lb) 0.048 Torr, and (c) 0.073 Torr. The E/Nthe external resistor was 510 1. was fixed at 14.5 Td.

4362 J. Appl. PhyS., Vol. 57, No. 9, 1 May 198£ W. C Wang and L. C. Lee 4362

t 1rzI . X PLLIA~% ILo V V J !\ -- % 4A' '1

20 stant. For these reasons, we limited our measurements toE/N less than 15 Td.

In order to check the H20 concentrations, as well as to

avoid the problem of impurities which may be dissolved inwater, the experiments were repeated using a premixed gas

4A mixture of 271 ppm H20 in Ar as the H20 source. TheE - 11.5 Td results for the electron attachment rate constants are the

- 5 FIG. 4. The v. /IH 201 values as a same as above.function of [H 20]/[Ar] for variousapplied EIN. B. H20-N2 mixture

-r\\ For the H20-N 2 mixture, the transient waveforms ob-

served with and without water vapor at E IN from 8 to 24Td do not show difference. In this E IN region, the mean

2.7 Td electron energy is in the range of 0.8-1.5 eV. 12.25 These1electron energies are too low for the dissociative attach-

0 O.4 08 t2 ment process,' 2"3 but too high for the nondissociative at-

I H20]/I Arl (10) tachment process.3"5 Therefore, one does not expect theelectron attachment process to occur.

rate constants increase with increasing E IN at a constant When E /Nis reduced, the waveforms with and without[H201/[Ar]. H20 in N2 show some differences. Two waveforms at E/

The extrapolated value of v. /[H20] at [H201 = 0, (v, N = 6.3 Td, one for 390 Torr of N2 without H20 and the

[H 20]), is associated with the electron energy distribution in other for 468 Torr of N2 with 2.7 Torr of H20 are shown inpure Ar. The electron attachment rate constants, (v. / Figs. 6(a) and 6(b), respectively, These pulses are represen-

[H20])o, as a function of E IN are shown in Fig. 5. No mea- tatives for several subsequent shots. The waveforms are

surements were made below E IN = 2.5 Td, because at lower very reproducible, and the difference between Figs. 6(a)

E/N the signal levels of V(t) were too low to measure the and 6(b) does have a statistical significance. As shown in

attachment rates with certainty. The onset for the electron Fig. 6(b), the pulse duration is shortened when H 20 is add-

attachment rate constant is estimated (from the data shown ed, indicating that the electron drift velocity increases as

in Fig. 5) to be at E/NI< 2 Td. This result is consisteit with H20 is added. We have measured the electron drift veloc-

the measurement of Hurst et al.5 that the onset is at E I ities at various (H201 and E IN, and found them Lo be con-

N = 1.2 Td. However, the data of Hurst et al.5 (in the E/ sistent with the published data. "

N = 1-3 Td region) do not join smoothly with the present Since the electron drift velocity increases with the addi-

data. A comparison is made in Sec. V. tion of H20, it is expected that the first peak in Fig. 6(b) willAt high E IN, the attachment rate constants do not inbe larger than that of Fig. 6(a). This expectation is different

crease as fast as those at low E IN as shown in Fig. 5. This from the waveforms observed, namely, the amplitude of the

may be accounted for by the fact that the mean electron first peak in Fig. 6(b) is smaller than that of Fig. 6(a). This

energy in Ar does not increase at high E IN. 2 In addition, result suggests that electron attachment occurs in this peri-

ionization may take place at high E IN which may interfere od. Furthermore, the amplitude of the second peak in Fig.with the measurement of the electron attachment rate con- 6(b) increases more than that of Fig. 6(a). This may result if

the electrons attached earlier are released later to enhancethe electron current at the end of the pulse. Comparing the

U, amplitudes of Figs. 6(a) and 6(b), the current due to detach-E, ment appears at a time less than 0.7 /is, indicating that the

* lifetime of this negative ion is less than this value.

10 00 0

oTime (ps)

0 2 40 0

U - 20.4/1

0 40- 0 * , , , , J , I , ,* a

W 0 5 10 15

-17 2 FIG. 6. The transient voltage waveforms produced from the electron mo-E / N (10 V-cm ) tion in (a) 390 Torr of N, without HO and (b) 468 Torr of N 2 with 2.7 Torr

FIG. 5. Electron attachment rateconstants obtained from the extrapolation of HO. The E /Nwas fixed at 6.3 Td. The electrode spacing was 3.7 cm, andof v./[H,01 at [H201 = 0 in the H20-Ar mixture as a function of E/N. the external resistor was 51012.


VA \~ k.,.7

Time (Ms)

0 0 0.4 0.8

08 0 ........., °'b,

S> 30

0.6 ' 601

S I IFIG. 8. The transient voltage waveforms produced from the electron mo-

0 1 2 3 tion in (a) 324 Torr of CH, without HzO and (b) 420 Torr of CH 4 with 3.6Torr of HO. The E/N was fixed at 5 Td, the electrode spacing was 3.7 cm,

H20 ] Torr) and the external resistor was 220 .(.

FIG. 7. The ratios for the magnitudes of the second peak to the first peak,V2/ V, as a function of (H2O] in N2 at E/N = 4.2 Td (0) and 6.3 Td (A). and 7.6 Td, respectively. For every case, In(V'/V) is linearly

dependent on t. The "apparent" electron attachment ratesmeasured from the slopes at various [H20] are shown in Fig.

Because of the first peak and the second peak being af- 11. At each E IN, the electron attachment rates increase lin-fected by the attachment and detachment process of the early with [H 201, from which "apparent" electron attach-temporary negative ion, the ratio of the amplitudes of these ment rate constants v. /[H 201 are obtained as shown in Fig.two peaks will give an indication for the attachment effect. 12. Note, the data were measured at several different CH,For an example, the ratios for the amplitudes of the second pressures. The attachment rate constants do not depend onpeak V2 to the first peak V, for the data taken at E IN = 4.2 the CH4 pressure. It is also worth noting that the attachmentand 6.3 Td at varied [H20] are shown in Fig. 7. The V2/V, is due to the formation of "temporary" negative ions, whichvalues do not depend on E IN at [H20] = 0, but it shows is different from the "stable" ones. We thus call the mea-

dependence as [H20] increased. The slope of the V2/V, for sured attachment rate as an "apparent" one.E IN = 4.2 Td is larger than that for E IN = 6.3 Td. This The second peak shown in Fig. 8(a) is significantly larg-suggests that the electron attachment rate for the forma- er than the middle portion, indicating that in CH4 the drifttion of temporary negative ion increases with decreasing velocity for electrons near the anode is significantly affectedE/N. This is consistent with the earlier observations3' 5 by the electrode. For the measurements ofattachment rates,that the electron attachment rate for the nondissociative the [H20] concentrations were kept as low as possible so thatattachment process increases with decreasing electron en- the electron drift velocities were not significantly alteredergy. This phenomenon is further studied by measuring the when H20 was added. For the amount of H20 (0.07-3.6electron attachment rate in the H20-CH4 mixture below. Torr) used in the experiment, the electron drift velocity at E /

We have repeated the above experiments by producing N = 7.6 Td is not significantly affected, which is reflected in

the initial electrons by two-photon ionization of a trace of Fig. 10(b) as the extrapolation of V7/V to t = Ois nearly equaltrimethylamine in N2. The results are consistent with the to 1. [The V'/Vat t = 0 is proportional to W'/Was shown inabove experiments. Eq. (4).1 At low E IN, the electron drift velocity depends

strongly on the [H20] added (see Fig. 9) so that the electronC. HgO-CH4 mixture

! 1 Two transient voltage waveforms atE/N = 6.3 Td, one

for 324 Torr of CH 4 without H20 and the other for 420Torr of CH4 with 3.6 Torr of H20 are shown in Figs. 8(a) t" "and 8(b), respectively. The pulse duration of 0.35 Ms here is *I.L' , E %°

much shorter than that in Ar of N2 gas. This is because the ,,,

electron drift velocity in CH4 is about one order of magni- 3

tude higher' 2 than that in Ar or N2. The drift velocities as a z .function ofE/Nat various water vapor contents in CH4 are -

shown in Fig. 9. It is noted that the electron drift velocity > "

becomes smaller when H20 is added to CH4, instead oflarger as H 20 is added to Ar of N2. The electron drift veloc- cities we measured agree well with the published data.'8

In Fig. 8(b), the second peak is much smaller than the ,ifirst peak [as compared with Fig. 8(a)], clearly indicatingthat an electron attachment due to H20 occurs in this short 01

0pulse period. The electron attachment rate can be measured20

from the ratio of the voltages with and without H20 in CH4.

The ratios of V'JV for the H20 pressures of 0.72, 2.16, and FIG. 9. The electron drift velocities in the H20-CH4 mixtures as a function

3.59 Torr are plotted in Figs. 10(a) and I0(b) for E IN 2.5 of E/N. The IH2O1/[CH4] are: (0)0%; (V) 0.4%; (U 1.2%; (A) 1.7%.


1 A C

> B>

CC

0.5 E

0.2 0.4 !0 5

Time ( Lds E/N (10-17V-Cm 2 )

FIG. 12. "Apparent" attachment rate constants at various E IN for theb H20-CH. mixture. The data were measured at the CH4 pressures of 325

(), 200 (A), and 135 U) Torr.A

mixture is short, the current induced by the detached elec-trons may become observable after the primary electrons are

0.5 absorbed by the anode. This expectation is indeed shown in'. . the transient pulses of Fig. 8(b), in which the "after-current"

0 02 04 of V'( )aftert = T'hasataillongerthanthatofV(t )showninTime (jus ) Fig. 8(a)aftert = T.The"after-currents"ofV(t)and V'(t)are

FIG. 10.The ratios of transient voltages measured with and without H 20 in plotted on a logarithmic scale as shown in Figs. 13(a) andCH4, V'/V, as a function of the elapsed time after the laser pulse at E/N of 13(b), respectively. The decay time in Fig. 13(a) is mainly the(a) 2.5 Td and (b) 7.6Td. Thegas pressures for H 20 in CH4 are: (A (0.72 Torr external RC time constant (R = 22012, C- 3 X 10-0 F). Onin 342 Torr; (B) 2.16 Tort in 379 Torr; (C) 3.59 Torr in 417 Torr. the other hand, the after current shown in Fig. 13(b) has an

additional long tail with a decay time of about 210 ns. Thisdrift velocity changes greatly with [H20]. However, the plot tail results from the detachment of the temporary ions,of ln(V'/ V) is still linear as a function of t as shown in Fig. whose lifetimes vary in the range of 190-220 ns as measured10(a). The W'/W values determined from the extrapolation at various [H20] and E IN.of V'/Vat t = 0 with a correction of laser power fluctuationare consistent with the electron drift velocities measuredfrom the pulse duration. This good agreement indicates thatthe present method is also applicable to electron attachmentrate measurement even for the case where the electron driftvelocity is changed by adding gases. 5

As shown in the measurements for the H20-N 2 mix- 57 nsture, the negative ion is a short-lived species, so that it will bedetached later. Since the transient pulse in the H20-CH4

74 ns

> /e00

c . . 0.1 LS0 0.1 0.2 (3

O0 -' Time( us)

0 2o

[.2c f. T.arr) FIG. 13. The amplitudes of after-currents in Figs. 81a) and 81b) as a functionof the time after primary electrons reaching the anode. The short decay

FIG. I I. "App.-re,'" .,,Iron attachment rate for [H,0] in (CH4] at E/Nof times for both curves (a) and (b) are caused by the external RC circuit. The(a) 2.5 Tel *) a;d fb .7.6 !*d (A). The CH4 pressures varied from 340 to 420 long decay time for (b) is caused by the electron detachment from the short-Torr as [4,0] in..rcased. lived negative ion.


Im

|~~~- . ,...fi r -. -- - - -- -7 - . r -, -r- w-r-w -, . . -t i; -. . . -- - - i - - - -

The electron detachment of the negative ions may affert

the measurement of the electron attachment rate so that the

measured attachment rate constant may appear low. How- 10-

ever, the transient pulse duration in CH4 (-400 ns) is not

nuch longer than the lifetime of the temporary negative ion, ,- H2 0 •also the concentration of negative ions is much less than that ,°of primary electrons, so that the effect of detachment on the E

measured attachment rate may not be serious. It is estimated

that the uncertainty of the attachment rate constant may not - •

be larger than + 30% of the given value.The earlier observations' " of electron attachment in _

H 20 at low electron energy were questioned 2"4 because nostable H2 0- ion was observed. 7 ' 2 ' 6 The attachment was

arbitrarily attributed'" 4 to the presence of impurities, mostlikely 02. We have examined this impurity problem careful-

ly. There are several observations to show that the attach-

ment observed in this experiment is not due to 02. First, the Lu

negative ion observed is short-lived, in contrast to the O2ion that is very stable. Second, the attachment is a two-body I I t I I I I ,,

process as evident from the fact that the attachment rate 01 1 0

constant is linearly dependent on [H20] (see Fig. 12). This is Electron Characteristic Energy ( ev)

quite different from the attachment process of 02 which is a FIG. 14. Electron attachment rate constants as a function of electron char-

three-body process so that one would expect the attachment acteristic energy in CH4, H20, and Ar. The data for H20 in CH4 (A) and Ar

rate to be quadratically dependent on the gas pressure. (V) were measured in this experiment. The data for pure H 20 ( (from

Third, the attachment rates measured are quite high. It is Ref. 14) and for H2O-Ar (*) (from Ref. 5) are plotted for comparison.

unlikely that any impurity has such high concentration and

high attachment cross section to make the attachment rate Hurst et al.'8 reproposed that the motion of low energyobservable. These facts clearly indicate that the attachment electrons in H may form a temporary negative ion. The

is not caused by the possible impurity, 02, but is caused by formation of temporary negative ions has been used to ex-

the formation of temporary ions. This point is further dis- plain' the decrease of electron drift velocity as HO is add-

cussed below. ed to CH 4 (see Fig. 9). The temporary ion of lifetime about200 ns observed in this experiment fits very well with the

V. DISCUSSION suggestion.H 2 0 is a strong electron-polar molecule, the cross sec-

The electron attachment rate constants are plotted ver- tion for the rotational excitation by low energy electrons is

sus the electron characteristic energies (D /u) in CH4 , H 2 0, very large (larger than 10- ,6 cm 2).3 4 3 5 The cross section for

and Ar as shown in Fig. 14. The electron characteristic the vibrational excitation of HO by low energy electrons

energy in CH, at each E/N is calculated from the data (<1 eV) was also observed to be very large (peak -

given by Cochran and Forester.3 The electron attachment 4 X 10-16 cm2).3 6 The large excitation cross section was at-

rate constant for pure H,O without buffer gas is obtained tributed to the dipole-dominated resonances. 6.1 [The

from the product of the electron attachment coefficient and H 2 0- (2A,) state calculated by Claydon et al. -" of about 2

the electron drift velocity at each E IN.'" The electron at- eV is too high to be involved in the observed attachment.] It

tachment coefficient for pure HO is the arithmetic mean is presumed that H 20 is excited by low energy electrons to

value given by various investigators" 13 2

.3 at each E/N as an intermediate compound state, (H20-)*, which then au-

reviewed by Gallagher et al. " The electron drift velocity todetaches to excited vibrational and rotational levels of

and the electron characteristic energy for pure H.0 at each the H 20 ground electronic state. The mechanism has been

E/N are also adopted from this review paper.' 4 The elec- described in detail by Schulz for the diatomic molecules.3 9

tron characteristic energy for Ar at each E/N is obtained The lifetime of a compound state is usually short (10- to

from a review paper given by Dutton. 5 The electron at- 10-- '" s), 9 and it is not expected to be observable in our

tachment rate constants measured by Hurst et al.' in Ar at experiment. However, H20 has a strong dipole moment, its

low E/N are also plotted for comparison, compound state may live long enough such that it can col-As stated before, the electron attachment in HO at low lide with other H,0 molecules to form a dimer

. elecl'-, energy reported before3 '" was subject to ques- H 20 • H2 0 or clusters H 2 0- (H20),. The ion cluster isti,!. I In order to resolve the discrepancy, Hurst et al.'7 generally more stable 0 than the compound state. The tem-

propose: a! icr that HO might quasitrap low energy elec- porary negative ions observed in this experiment may be

trons, for which the "quasitrap" H,O may have a limited dimers or clusters. Therefore, the lifetime observed may

lifetime in the order of 10- 1-10 7 s. Later, in order to belong to the clusters instead of H 20-. The measured elec-

explain the measured momentum transfer cross section for tron attachment rates for the formation of temporary nega-

H 20 being larger than the theoretical value by a factor of 2, tive ions depend linearly on [H20] (see Fig. 11) and not on

4366 J. Appi. Phys., Vol. 57, No. 9, 1 May 1985 W. C. Wang and L. C. Lee 4366

* - ~ ~ vr~'.%V R,~ . 'v ? * V

[CH4] (see Fig. 12). This indicates that the rate for the for- fessor M. A. Fineman at San Diego State University, andmation of clusters from the (H20-)* compound state may Dr. E. R. Manzanares, Dr. J. B. Nee, and Dr. M. Suto inbe very fast so that the primary electron attachment to H20 our laboratory for useful suggestions and discussion. Thisis the rate controlling process. work is supported by the Air Force Office of Scientific Re-

For high energy electrons, the attachment is due to a search, Air Force Systems Command, USAF, under Grantdissociative attachment process. The attachment rate con- No. AFOSR-82-0314.stants measured in Ar indicate that the attachment beginsat a mean electron energy of about 4 eV. This is consistent 'K. Schoenbach, G. Schaefer, M. Kristiansen, L. L. Hatfield, and A. H.

with the thresholds 12.13 for producing 0- and H- from Guenther, Electrical Breakdown and Discharges in Gases, Part B, edited

dissociative attachment of H20 (i.e., 4.9 and 5.5 eV, respec- by E. E. Kunhardt and L. H. Luessen (Plenum, New York, 1983), p. 415.tively). The attachment rate constants reported by Hurst et 2j. K. Burton, D. Conte, R. D. Ford, W. H. Lupton, V. E. Scherrer, and 1.

M. Vitkovitsky, Digest of Technical Papers, 2nd IEEE InternationalaL. are higher than the present values (see Fig. 14). The Pulsed Power Conference, Lubbock, TX, edited by A. Guenther and M.

reason for this discrepancy is not known. The dissociative Kristiansen, (IEEE, New York, 1979), p. 284.

attachment process has been used to explain the observa- 'E. Kuffel, Proc. Phys. Soc. London 74, 297 (1959).

tion that the addition of water vapor to dry air increases the 'A. N. Prasad and J. D. Craggs, Proc. Phys. Soc. London 76, 223 (1960).'G. S. Hurst, L. B. O'Kelly, and T. E. Bortner, Phys. Rev. 123, 1715 (1961).

breakdown voltage. 6R. W. Crompton, J. A. Rees, and R. L. Jory, Aust. J. Phys. 18, 541 (1965).

The magnitude of the attachment rate constant for elec- 7J. L. Moruzzi and A. V. Phelps, J. Chem. Phys. 45,4617 (1966); J. L. Pack

trons in pure water is about the same as that in Ar as shown and A. V. Phelps, J. Chem. Phys. 45,4316 (1966).eJ. E. Parr, and J. L. Moruzzi, J. Phys. D 5, 514 (1972).

in Fig. 14, indicating that the electron attachment in H 20 is 9C. E. Klots and R. N. Compton, J. Chem. Phys. 69, 1644 (1978).

due to the dissociative attachment process. However, the ,oJ. L. Pack, R. E. Voshall, and A. V. Phelps, Phys. Rev. 127,2084 (1962).mean electron energy in H20 in Fig. 14 is less than the "l.S.Buchel'nikova,Zh.Eksp.Teor.Fiz.35,1119(1958)[Sov.Phys.

JETP35, 783 (1959)].threshold for the O or H- formation. This result suggests 'L. G. Christophorou, Atomic and Molecular Radiation Physics (Wiley,

that the electron energy distribution in H,O may be broad, New York, 1971).and/or the mean electron energy is not well measured. Be- 1H. Massey, Negative Ions (Cambridge U. P., London, 1976), p. 347.

cause the electron attachment rate constants are nearly the "J. W. Gallagher, E. C. Beaty. J. Dutton, and L. C. Pitchford, J. Phys.same for both HO-Ar and pure HO, it is likely that the ,5Chem. Ref. Data 12, 109 (1983).

". E. Bradbury and H. E. Tatel, J. Chem. Phys. 2, 835 (1934).electron energy in H20 is not much broader than Ar, al- "J. F. Wilson, F. J. Davis, D. R. Nelson, R. N. Compton, and D. H. Crawd-though the possibility of a wide energy spread in H 20 is not ford, J. Chem. Phys. 62, 4204(1975).ruled out. The attachment at an electron energy lower than 7G. S. Hurst, L. B. O'Kelly, and J. A. Stockdale, Nature 195, 66(1962).

"G. S. Hurst, J. S. Stockdale, and L. B. O'Kelly, J. Chem. Phys. 38, 2572threshold may suggest that the published data for the mean (1963)."electron energy" in H 20 are too low. The mean electron 'L. C. Lee and F. Li, J. Appl. Phys. 56, 3169 (1984).energy in H 20 has not been studied too thoroughly to date.' 4 20H. F. A. Verhaart and P. C. T. Van der Laan, J. Appl. Phys. 53, 1430

(1982); 55, 3286 (1984).2 'L. G. Christophorou, J. G. Carter, and D. V. Maxey, J. Chem r iys. 76,2653(1982).

VI. CONCLUDING REMARKS 22Th. Aschwanden, Gaseous Dielectrics IV, Proc. 4th Int. Symp. Gaseous

We have applied a different approach to investigate the Dielectrics, edited by L. G. Christophorou (Pergamon, New York, 1984),pp. 24-33.

transient electron conduction current in the gas mixture of 23L. G. H. Huxley and R. W. Crompton, The Diffusion and Drift of Elec-

a trace of H20 in Ar, N2, and CH 4. This method is different trons in Gases (Wiley, New York, 1974), pp. 298-303.from the conventional swarm method in which negative 24F. Li and L. C. Lee, unpublished.

ions are detected. We have demonstrated in this paper and "J Dutton, J. Phys. Chem. Ref. Data 4, 577 (1975).a E. B. Wagner, F. J. Davis, and G. S. Hurst, J. Chem. Phys. 47,3138 (1967).

a previous paper 9 that the present method is useful for 2 L. C. Lee and W. K. Bischel, J. Appl. Phys. 53, 203 (1982).measuring the "apparent" electron attachment rate con- "L. C. Lee and F. Li, unpublished.stants due to the formation of short-lived species. 2 D. D. Errett, "The Drift Velocity of Electrons in Gases," thesis, Purdue

Foran opening switch, the electron attachment rate dur- University, W. Lafayette, IN (1981).F10R. A. Nielsen, Phys. Rev. 50, 950 (1936).

ing the opening period needs to be high, that is, the electron "L. W. Cochran and D. W. Forester, Phys. Rev. 126, 1785 (1962).

attachment rate is required to increase with increasing El "H. Ryzko, Ark. Fys. 32, 1 (1966).N. The results of our experiment show that the H20-Ar "A. V. Risbud and M. S. Naidu. J. Phys. (Paris) Colloq. C7 40, 77 (1979).

mxuehsY. Itikawa, . Phys. Soc. Jpn. 32, 217 (1972).mixture has this characteristic. The increase of the electron "K. lung, Th. Antoni, R. Muller, K. H. Kochem, and H. Ehrhardt, J.drift velocity due to the addition of H20 to Ar is an addi- Phys. B 15, 3535 (1982).

tional benefit for the design of opening switch, because the "G. Seng and F. Linder, J. Phys. B 9, 2539 (1976).electron conduction current will be increased during the "S. F Wong and G. J. Schulz, Proceedings ofthe 9th International Confer-

ence on Physics of Electronic and Atomic Collisions (University of Wash-switching period. ington, Seattle. 1975).

'"C. R Claydon. G. A. Segal, and H. S. Taylor, J. Chem. Phys. 54, 3799ACKNOWLEDGMENTS (197 I1.

"G. J. Schulz, Rev. Mod. Phys. 45, 423 (1973).The authors wish to thank Professor K. Schoenbach and "'S. G. l.ias and P. Ausloos, Ion-Molecule Reactions (American Chemical

Professor G. Schaefer at the Texas Tech University, Pro- Society, Washington, DC, 1975), pp. 165-197.

4367 J. Appl. Phys, Vol 57, No. 9. 1 May 1985 W C Wang and L C Lee 4367

N ~

S..

5 ,

Appendix C

"Shortening of Electron Conduction Pulsesby Electron Attacher C3F8 in Ar, N2 and CH4"

nw-

r..

I

3r~~~~r'r~~- W' W1 't _. U ,. - 7 - r- -a 7~ '

Shortening of electron conduction pulses by electron attacherCF 6 in Ar, N2, and CH 4

W. C. Wang and L. C. LeeDepartment of Electrical and Computer Engineering, San Diego State University, San Diego,California 92182

(Received 19 February 1985; accepted for publication 19 March 1985)

The attachment rate constants of C3 F, in the buffer gases of Ar, N2, and CH4 are measured using aparallel-plate drift-tube apparatus. The dependences of the electron drift velocities on thecontents of C3Fs in various buffer gases are investigated. Electrons are produced by irradiating thecathode with ArF laser photons. The transient voltage pulses induced by the electron motion areobserved. We find that the C3F,-CH4 mixture has the desirable characteristics of both electrondrift velocity and attachment rate constant for the application of diffuse-discharge openingswitches.

I. INTRODUCTION across a resistor connecting the anode to ground. The tran-

It has been pointed out that fast opening switches are sient voltage was monitored by a storage oscilloscope (HP

needed'"2 for developing a magnetic energy storage system. 3 model 1727A) and photographed for a permanent record.For the design of a fast opening switch, the duration of the The electron attachment rates are determined by comparing

electron conduction pulse and the electron drift velocity are the transient voltages with and withouthC in various buff-

the important parameters. The pulse duration can be shorter gases as described in earlier papers.a ° The electron drift

ened by attaching electrons to electronegative gases in var- velocities were determined from the duration of transient

ious buffer gases. A gas mixture suitable for a diffuse-dis- pulses and the electrode separation.

charge opening switch has to fulfill the requirements" 2'4: (i) The gas pressure in the cell was maintained constant in

high electron drift velocity and low electron attachment rate a slow flow system (flow rate - 20 STP cm3/min). The gas

at low density-reduced electric field (EIN- 3 10-17 pressure was measured by an MKS Baratron manometer.

V cm2), and (ii) low electron drift velocity and high electron All measurements were done at room temperature. The Ar,

attachment rate at high E IN. In our research program, we N2, and CH4 gases (supplied by MG Scientific) have mini-

have developed a method to measure the electron attach- mum purities of 99.998%, 99.998%, and 99.99%, respec-

ment rates and electron drift velocities of various gas mix- tiv;'ly. The perfluoropropane (C3F,) supplied by Matheson

tures for the application of opening switches. has a minimum purity of 99.0%. Because the C3F, concen-

Recently, Christophorou's group ' have measured the trations used in the Ar experiment were usually small, a di-

electron attachment rates and electron drift velocities for luted 3% of CF, in prepurified Ar (supplied by Matheson)

various binary gas mixtures by a swarm method. Among was used so that the partial pressure of C3F could be mea-

these gas mixtures, C.F,-Ar and C3F-CH4 were suggested sured accurately.

as the good candidates for diffuse-discharge opening switch- III. RESULTSes. In this paper we report the electron transport parametersof these mixtures using a different method. [We have already A. CF-Ar mixture

applied this method to measure the electron attachment The electron transient pulses for C3Fs-Ar mixtures atrates of various gas mixtures such as N20-N 2 (Ref. 9) and E/N= 3 Td (1 Td = 10- 17 V cm2) are shown in Fig. 1,H20-Ar (Ref. 10).] From our results in this study, we found where the pressure of Ar is 380 Torr, and the pressures ofthat in the case of low CF 8 concentration the C3F-CH, andC3F8-N2 mixtures are suitable for the application of openingswitches, but the C3F,-Ar mixture is not. Time (As)

II. EXPERIMENT 0 0 6 12

The experimental apparatus and the technique for data Z- -- - --- - -.9o > 20. 1,

analysis have been described in detail in previous papers.9 .m E

Briefly, the gas cell is a 6 in. six-way aluminum cross. Theelectrodes are two parallel uncoated stainless steel plates of 5 > 40.

cm in diameter and 3.7 cm apart. Electrons were producedby irradiating the cathode with ArF laser (193 nm) photons(Lumonics model 861 S). The pulse duration is about 10 ns. A FIG. I. The transient voltage waveforms produced from the electron mo-negative high voltage was applied to the cathode. The con- tion in 380 Torr of Ar (a) without C3F. (b) with 0.02 Torr, and (c) with 0.04

Tort of CF.. The electrons were produced from irradiation of the cathodeduction current induced by electron motion between the by ArF laser photons. TheE/Nwasfixedat3Td, the electrode spacing waselectrodes was converted into a transient voltage pulse 3.7 cm, and the external resistor was 2 kfl.

184 J. Appl. Phys. 58 (1), 1 July 1985 0021-8979/85/130184-04$02.40 ® 1985 American Institute of Physics 184

10(a) Ar (c) CH4 •

S2 Q

A

I -b) N E

*6

S 0.1

C V I

uj

14.

0.0 I n i I ,VI

0 10 20 30 40

E/N 1lo17 v-cm 2 )0.2 . . . . i . .0 10 20

FIG. 2. Electron attachment rate constantsof C3F. in the buffer gasesof (a) E/ N (10"17 V-cm2]

380 Tort of Ar, (b) 320 Torr of N 2, and (c) 250 Torr of CH,.FIG. 4. The electron drift velocities in the C3F.-N 2 mixture as a function of

CF, are (a) 0, (b), 0.02, and (c) 0.04 Torr. As shown in the E/IN. The [C3F,]/[N.] ratios are 0 (A), 0.63% (0), and 1.5% (V).

figure, the electron pulses are shortened when smallamounts Of CAF are added to At. The shortening is caused long time to open a discharge switch,."' Therefore, for theby the electron attachment of CF. The electron attachment case of low CF, concentration the CF-Ar mixture is notrate v,, for a CAF concentration is obtained from the ratio of desirable for the application of diffuse-discharge openingtransient voltages with and without CAF in Ar as a function switches. However, in a practical switching device, the CF,of time (see previous paprs?"1 for detail). The eecron a- concentration can be so high that the CAF-Ar mixture istachment rate constant k. is determined by y.o/[CF]. We useful for the switching application as stated in Ref. 5.find that the k. values decrease with increasing [CF]/[Ar], The electron drift velocity Wis measured in this experi-as reported earlier by Hunter and Christophorou.' This is ment, which is equal to the ratio of the electrode spacing tolikely caused by the effect Of CAF on the electron energy the electron drift time. Because of the fast decay of conduc-distribution in At, similar to the case"° " of H20 in At. The tion pulse at high ICYs] (see Fig. ), the electron drift veloc-attachment rate constant of CFs is the value of extrapolat- ity is only measurable at low CAF concentration ( < 0.02%)ing Vo /[CF8] to ICAF] = 0, for which the electron energy in Ar. The measured electron drift velocity increases sub-distribution is associated with pure At. The results of k. at stantially when a small amount Of CAF is added to Ar. Atvarious E/IN are shown in Fig. 2(a), where the Ar pressure is high E/IN, the electron drift velocity for the C F-Ar mixtureabout 380 Ton'. The experimental uncertainty is estimated merges with the value for pure Ar. Our observation for theto be ±- 20% of the given value. [Figs. 2(b) and 2(c) show the electron drift velocity of the CAF mixture is consistent withelectron attachment constant for the CFs-N2 and CAF- the measurements of Christophorou et al.4 'CH4 mixtures as discussed in Sec. III B and Sec. III C.]

The attachment rate constant is about 6 X 10- no cm3/ A B. CF-Na mixtureat E/N = 1.2 Td, and decreases with increasing E/N. A gas

mitrihtentueo hsEI epnec iltk The electron conduction pulses for 0, 0.2, and 0.8 TorrSOf CA in 320 Torr of N2 are shown in Figs. 3(a), 3(b), and

STime u s ) 3(c), respectively, where the E/INis fixed at 8.9 Td. Again,

0 2 4 the shortening of electron pulses in Fig. 3 is due to the elec-01',.._.-:-/. tron attachment to CAF. The electron attachment rate con-

SC l ... /stants ofC3F, in N2 are slightly dependent on the contents of

4,

" 4L0.| / " CAF as reported earlier by Hunter and Christophorou." TheEk values measured at [C3Fw] - 0 are shown in Fig. 2(b) as a

> 80, / function of E/IN from 5 to 25 Td. The attachment rate con-] stants of C3F8 in N2 increase with increasing EIN. This char-

acteristic is desirable for the application of a diffuse-dis-FIG. 3. The transient voltage waveforms produced from the electron too- charge opening switch.tion in 320 Tort of N2 (a) without CF,, (b) with 0. 2 Torr, and (c) with 0.8Torr of CjFx. The E/IN was fixed at 8.9 Td. The electrode spacing was 3.7 The electron drift velocities measured at various E /Nascm, and the external resistor was 5 10 D/. various [CF,]/[N,] ratios are shown in Fig. 4. When a small

185 J. Appl. Phys., Vol. 58, No. 1, 1 ,July 1985 W. C. Wang and L. C. Lee 185

0.0, I -..I0~- I0N0304

Time ( us) 10

0 0. 0.8

cm n teetenlreitr -. 1 .-0

120 C

FIG. 5. The transient voltage waveforms produced from the electron mo-tion in 250 Torr of CH. (a) without C3F., (b) with 0. 15 Tort, and (c) with 1.0Tort of CFs. The E/IN was fixed at 39 Td. The electrode spacing was 3.7 Ecm, and the external resistor was 51 D. .

small amount of CFs is added to N 2, the drift velocity in- 0

creases at low E IN but remains unchanged at high E IN, _which is similar to the characteristic observed in the C3F8 - LUAr mixture. ~ The electron drift velocity shows a small bump

when more CF, is added (see Fig. 4). The increase of elec- 0.3 0.5 1 5 10tron drift velocity due to the addition of C3F to N2 is also acharacteristic desirable for the design of diffuse-discharge Mean Electron Energy (ev)

opening switches. FIG. 7. Electron attachment rate constants of C3Fs in Ar (0) and N2 (v) asa function of mean electron energy. The data from Ref. 5 (solid line) are

C. CaFa-CH4 mixture plotted for comparison.

The electron conduction pulses with 0, 0.15, and 1.0Torr of CFg in 250 Torr of CH4 are shown in Figs. 5(a), 5(b), Ar and N2, the pulse duration observed here is relativelyand 5(c), respectively, where the E/Nis fixed at 39 Td. Since r The pule rationsobserved here is reathe eletshort. The attachment rate constants of C3F in CH mea-the electron drift velocity in CH4 is much higher than that in sured at various E IN are shown in Fig. 2(c). The k. values

increase by a factor of about 40 from E IN = 5 to 39 Td asshown in Fig. 2(c). The dependence of this electron attach-ment rate constant onE IN is very desirable for the applica-

!. tion of diffuse-discharge opening switches.

The electron drift velocities for the CFs-CH4 mixturemeasured at various [C3F,]/[CH,] ratios as a function of

1.O ro. E IN are shown in Fig. 6. The experimental uncertainty for~the electron drift velocity is estimated to be within ± 10% ofthe given value, which is mainly caused by the time measure-

, ",ments. The uncertainty for the time measurement in the be-CD ginning or the end of a pulse is about 20 ns. This causes an

-'"uncertainty for the electron drift velocity of about 7%. The"" "T - measurements of Hunter et al.' with [C3F8]/[CH,] = 0.5%

o ""- ...... .and 2%, and the measurement of Hurst et al.'" with pure

. CH, are also plotted in Fig. 6 for comparison. Our measuredvalues are between those two previous results and are consis-

tent within experimental uncertainties. The electron drift ve-

locity is affected by the addition of C3F8 to CH4. The elec-t; tron drift velocities have maxima in the E/N region of 5-10"W Td. The peak position shifts toward the high E IN side when

the CFs concentration increases. Comparing with the C3Fs-Ar and C3Fs-N 2 mixtures, the electron drift velocity in the

0 1 C3 F,-CH4 mixture is less sensitive to the contents of CF,.0 10 20 30 40 The electron drift velocity for the CF 5 -CH, mixture is very

E / N ( i0 17 V-cm 2 ) high (- 1. 1 X 107cm/s)at moderateEIN (5-7Td). Althoughthe electron drift velocity decreases with increasing E IN, its

FIG. 6. The electron drift velocities in the CF 1 -CH4 mixture as a function value at E IN = 40 Td is still quite large (6 X 106 cm/). ThisorE/N. The [CF.]/[CH,] ratios areO (A), 0.48% (0), and 1.79% (Y). Thedata from Ref. 8 with [C F,]/[CH,] = 0.5% (- -) and 2% (--), and from characteristic is very desirable for designing diffuse-dis-Ref. 12 with pure CH4 (. .. ) are plotted for comparison, charge opening switches.

186 J. Appl. Phys., Vol. 58, No. 1.1 July 1985 W. C. Wang and L. C. Lee 186

IV. DISCUSSION AND CONCLUSION that this dissociative electron attachment may be a major

For a comparison of our measurements with the pub- process occurring at high E IN.lished data, the attachment rate constants of CF, in N2 and In summary, we have employed a relatively new meth-

Ar are plotted in Fig. 7 as a function of the mean electron od to measure the electron attachment rate constants of

energy (e), where [Ar] = 380 Torr and [N2 ] = 320 Torr. C3F, in the buffer gases of Ar, N, and CH4. Our measuredThe mean electron energies in N2 and Ar at each E IN are k. values as a function of (e) are consistent with the pub-obtained from Ref. 5. The data of Hunter and Christo- lished data." In this study, we find that the C3F,-CH4 mix-

phorou are plotted as the solid line in Fig. 7, where the ture has a high electron drift velocity and its electron attach-

buffer gas pressure (N, or Ar) is 1000 Torr. ment rate constant increases with increasing EIN. TheseChristophorou's group-5* 3 have observed that the elec- characteristics are desirable for the application of diffuse dis-

tron attachment rate constant of C3F, depends on the total charge opening switches.

pressure for pressures larger than 1000 Torr. This pressure ACKNOWLEDGMENTdependence was attributed to the processes5

C3F + e, (1) The authors wish to thank Dr. E. R. Manzanares, Dr. J.B. Nee, Dr. M. Suto, and Dr. D. Taylor in our laboratory,

CF + e + C3F8 and Professor M. A. Fineman at San Diego State Universityfor useful suggestions and discussion. This work is support-

C3F8 + energy, (2) ed by the Air Force Office of Scientific Research, Air ForceOffice Systems Command, USAF, under Grant No.

where an electron initially captured by C3F8 forms a tempo- AFOSR-82-0314.rary parent negative ion C3F, - *. The parent negative ionwill either autoionize or be stabilized by collision with the 'K. H. Schoenbach, G. Schaefer, M. Kristiansen, L. L. Hatfield and A. H.

Guenther, IEEE Trans. Plasma Sci., PS.10, 9246 (1982).buffer gas. So the higher the buffer gas pressure is, the larger 'K. H. Schoenbach, G. Schaefer, M. Kristiansen, L. L. Hatfield, and A. H.the electron attachment rate will be. Guenther, in Electrical Breakdown and Discharges in Gases, Part B, edited

In our experiment, we have measured the electron at- by E. E. Kunhardt and L. H. Luessen (Plenum, New York, 1983), p. 415.tachment rate constants as a function of the buffer gas pres- 3J. K. Burton, D. Conte, R. D. Ford, W. H. Lupton, V. E. Scherrer, and I.sure for pressures less than 1 atmosphere. The measured M. Vitkovitsky, in Digest of Technical Papers, 2nd IEEE International

Pulse Power Conference, Lubbock, Texas, edited by A. Guenther and M.electron attachment rate constants are not sensitive to the Kristiansen (IEEE, New York, 1979), p. 284.buffer gas pressure. For instance, for the C3 F-Ar gas mix- "L. G. Christophorou, S. R. Hunter, J. G. Carter, and R. A. Mathis, Appl.

ture at E/N = 2.2 Td ((e) = 3.55 eV), the electron attach- Phys. Lett. 41, 147 (1982).'S. R. Hunter and L. G. Christophorou, J. Chem. Phys. S0, 6150 19941.mert rate constants increase less than 100/o when the Ar 'L. G. Christophorou and S. R. Hunter, Electron-Molecule Interactions

pressures increase from 270 to 760 Torr. The relative less and Their Application, Vol. 2, edited by L. G. Christophorou (Academic,

dependence of the electron attachment rate constants on the Orlando, 1984), p. 317.

buffer gas pressure in this experiment may be due to the high 'L. G. Christophorou, S. R. Hunter, J. G. Carter, S. M. Spyrou, and V. K.Lakdawala, Proceeding of the 4th IEEE International Pulsed Power Con-

mean electron energy we studied (0.8-6 eV) in which the ference, edited by T. H. Martin and M. F. Rose (Texas Tech., Lubbock,dissociative electron attachment process is important. The TX, 1983). p. 702.

dissociative electror, attachment processes of CF, are given 'S. R. Hunter, J. G. Carter, L. G. Christophorou, and V. K. Lakawala,

as 14 Gaseous Dielectric IV, edited by L. G. Christophorou and M. 0. Pace (Per-gamon, New York, 1984), p. 224.

C3F8 -+ e - C. Fy- +C 3 _Fs_ (3) 9L. C. Lee and F. Li, J. Appl. Phys. 56, 3169 (1984).'"W. C. Wang and L. C. Lee, J. Appl. Phys. 57,4360 (1985)."G. S. Hurst, L. B. O'Kelly, and T. E. Bortner, Phys. Rev. 123, 1715)(1961).where x = 0-3 and y = 0, 3, 5, 7. The lowest appearance "G. S. Hurst, J. A. Stockdale, and L. B. O'Kelly, J. Chem. Phys. 38, 2572

onset energyt 4 for these processes is 1.1 ± 0.1 eV that pro- (1963.duces C2F3 -. The electron attachment due to this dissocia- "S. R. Hunter, L. G. Christophorou, D. R. James. and R. A. Mathis, Gase-tive attachment process will show pressure independence. ous Dielectrics III, edited by L. G. Christophorou (Pergamon, New York,

1982), p. 7.Our observation that the measured electron attachment rate "S. M. Spyrou, I. Sauers, and L. G. Christophorou, J. Chem. Phys. 78, 7200

constants are not sensitive to the buffer pressure indicates (1983).

187 J. Appl. Phys., Vol. 58, No. 1, 1 July 1985 W. C. Wang and L. C. Lee 187

FILMED

DTIC ~~2j

S 32! L

ATTACHMENT RND CHARGE I/i PROCESS DEPT OF ELECTRICAL …

Documents

Transcript of ATTACHMENT RND CHARGE I/i PROCESS DEPT OF ELECTRICAL …